“Almost all jobs involve tasks that AI can transform”
The following article is from the Stanford Graduate School of Business portal and was written by team member Lisa Bonos.
Economist Erik Brynjolfsson asserts that the AI-driven transformation of work is only just beginning, and that our institutions are not prepared for it.
“Although most workers do not yet use AI, they soon will,” notes Erik Brynjolfsson.
In brief
Economist Erik Brynjolfsson studies how AI can complement human work rather than replace it.
He suggests focusing on tasks, not jobs: “Almost no job is fully automatable, whereas almost all jobs include tasks that AI can transform.”
Brynjolfsson has urged tech and political leaders to act quickly to define the economic impact of AI.
How will advances in artificial intelligence transform jobs, companies, and the economy at large? What can leaders and employees do to prepare for a world reshaped by AI?
Erik Brynjolfsson—a professor (by courtesy) of Economics and of Operations, Information and Technology at the Stanford Graduate School of Business, as well as director of the Stanford Digital Economy Lab—has spent years researching these questions and anticipating the transformative effects of new technologies.
Among his many recent projects, he has closely monitored how companies implement AI, identified which workers stand to benefit most from it, and called on economists, AI researchers, and policymakers to act now “to create the incentives, safeguards, and institutions needed to steer AI in a direction that complements humans and benefits society.”
Brynjolfsson spoke with *Insights* about what it takes to successfully reimagine companies and institutions, thereby enabling humans to collaborate with AI to boost productivity.
Based on your research, how can AI be implemented while also considering its impact on real people? Erik Brynjolfsson: At the Digital Economy Lab, Elisa Pereira, Alvin Wang Graylin, and I analyzed 51 AI implementations that generated value across various sectors, such as manufacturing, financial services, and technology. We discovered that 77% of the toughest challenges were not technical in nature but related to change management, data quality, and process redesign. The key to success rarely lay in the AI model itself, but rather in the organization. The same technology led to vastly different outcomes. Workforce reduction was a significant factor, but only in 45% of cases; in the others, workers were reassigned to higher-value tasks.
How do you arrive at the best answer? By thinking in terms of tasks, not jobs. Almost no job is fully automatable, yet virtually all of them include tasks that AI can transform. When work is broken down this way, the capacity that is freed up creates the possibility of reassigning workers rather than cutting the workforce.
One of your indicators, the *Canaries Dashboard*, shows a much sharper drop in employment among young people aged 22 to 25 in AI-exposed occupations compared to older workers. How do you think entry-level jobs are changing?
Erik Brynjolfsson: In a recent study conducted with Bharat Chandar and Ruyu Chen, we observed a 19% decline in employment among 22- to 25-year-olds in occupations most exposed to AI, compared to their counterparts in less exposed roles. However, the most interesting finding is that this drop was greatest in occupations where AI automates human work rather than complementing it. The routine aspects of entry-level work—initial drafts, basic inquiries, preliminary screening—are precisely what these systems excel at. What remains is judgment, verification, client interaction, and AI coordination. Therefore, for many, the path to taking the first steps in their professional careers becomes more of an uphill battle.
What advice do you have for students entering the workforce?
Erik Brynjolfsson: My advice is to collaborate with people who are different from you and focus on making something actually work.
Last spring, I taught a course on economics and computer science titled *The AI Awakening*. The class was composed of an equal mix of GSB students and engineering or computer science students, and featured guest speakers from leading AI labs, venture capital firms, and cutting-edge startups. Three-person teams had 10 weeks to develop not just a business plan, but also a functional prototype capable of autonomous action (an “agentic prototype”). Students learned from one another—especially when coming from different disciplines—and created something tangible.
If mastering AI isn’t enough to protect your job, what will?
Erik Brynjolfsson: Technical mastery is important, but insufficient. Mastering AI in 2026 will be like mastering email in 2005. What truly protects you is knowing how to ask the right questions to guide your work with AI. Another key factor is taking responsibility for the outcomes; that fosters relationships and trust. In those situations, the machine augments your value rather than replacing you.
The most practical approach is to assess: which tasks does AI replace, and which does it enhance? You should then focus your efforts on the latter group before someone else makes that decision for you. It’s not a skill you acquire, but a habit: the habit of knowing how to reallocate your time.
How should policymakers and technologists prepare for the seismic shifts AI will bring?
Erik Brynjolfsson: I recently helped draft a joint statement on AI and transformation alongside other economists specializing in technology—Ajay Agrawal, Anton Korinek, and Tom Cunningham. It was striking to see the broad consensus among so many economists—who often disagree on other issues—regarding the impact of AI.
First, we need to measure the impacts. Second, stop making predictions and start building. Institutions take years to build, and transformative AI might not wait for them to be ready. Third, we need to make the complementary path the most profitable one. This means fiscal neutrality between labor and capital, and collaborations that reward the joint performance of humans and machines.
What do most workers who barely use AI understand that developers don’t see?
Erik Brynjolfsson: A Pew study reveals that about 65% of U.S. workers say they use AI little or not at all. When new general-purpose technologies like AI emerge, years of investment in intangible assets can pass before quantifiable gains materialize. My colleagues and I call this dynamic the “productivity J-curve.” We are currently at the point on the J-curve where initial investments haven’t yet yielded major breakthroughs; therefore, even though most workers aren’t using AI yet, they soon will be.
Our research on how AI is implemented illustrates this concretely. An executive told us that the hardest part of the job lies in process documentation and data architecture. Only 6% of the companies we analyzed had clean data to start with. The people doing the work already knew that.
In 2036, what indicator would you look at to see if we got it right?
Erik Brynjolfsson: GDP-B, a new measure of well-being and growth that I devised with colleagues at Stanford; it accounts for how much consumers benefit from goods and services, not just how much they pay for them. According to our estimates, U.S. consumer surplus derived from generative AI amounts to about $172 billion this year. Basically, none of this is reflected in traditional GDP measurements.
GDP was one of the great inventions of the 20th century, but today it leaves out an increasingly large share of the value generated in the 21st century. We need a new metric, such as GDP-B.
Companies Succeeding with AI Are Not Replacing Their Workers
This article originally appeared on Forbes and was written by contributor Gary Drenik. Gary Drenik writes about AI, data analytics, and innovation.
Summary
New research warns companies against using AI primarily to downsize their workforces, arguing that this “status quo bias” jeopardizes long-term growth by undermining crucial human capabilities such as creativity and judgment. Instead, experts advocate for using AI as a tool to augment capabilities—expanding employee potential and fostering innovation.
Surveys of workers reveal significant concerns:
31% fear losing their jobs and 25% distrust AI, highlighting the need to build trust and involve staff in its adoption. An automation-first approach can deplete the leadership pipeline and lead to “workslop” (substance-free AI-generated content that increases the human workload). Truly successful AI strategies will prioritize improving work quality, employee decision-making, and leadership development, transforming AI into a lasting competitive advantage by empowering people rather than simply replacing them.
In a complex business environment, artificial intelligence can seem like an obvious answer for executives seeking to streamline operations, cut costs, and achieve new efficiencies. However, as AI becomes integrated into daily business activities, the critical question is whether those benefits will endure.
New research led by Professor Jan-Emmanuel De Neve of the University of Oxford’s Saïd Business School
suggests that companies may be making a strategic error if they use AI primarily to cut staff. The study argues that companies overly focused on automation risk weakening the very human capabilities they need for growth—such as creativity, judgment, institutional knowledge, and leadership development. “It is a classic manifestation of status quo bias,” states De Neve. “For an executive, it is far easier to look at a spreadsheet and envision using AI to optimize tasks staff already perform than to undertake the arduous strategic work of rethinking how technology can create entirely new value. Viewing AI solely as a tool for workforce reduction may yield quick, visible cost savings; however, in the long run, it leads to organizational contraction that sacrifices the company’s capacity for innovation.”
This distinction is fundamental to AI strategy. Automation replaces tasks currently performed by people. Augmentation, on the other hand, expands the scope of what people can do, empowering employees to solve problems, manage customer relationships, and make higher-value decisions.
A recent survey conducted by my company, Prosper Insights & Analytics, illustrates the importance of this aspect.
In the study of U.S. adults, executives and business owners were more likely to report using generative AI (53%) compared to employees (38%). This suggests that leaders have a closer grasp of AI’s strategic potential, whereas employees experience the disruption it causes in their daily work more directly.
Prosper Insights & Analytics also revealed that 31% of employees worry that AI will lead to job losses, while 11% are concerned about losing their own jobs due to the technology. Additionally, 25% of employees state they do not trust AI to look out for their best interests.
These figures are significant because AI adoption involves an element of trust. If employees perceive that AI is being introduced to facilitate their replacement, they are less likely to openly experiment with it, question its outputs, or point out areas to management where the technology could improve the business.
Professor Ashish Kumar of Trinity Business School calls this an “agency and incentive problem.”
Senior executives often view AI through the lens of strategy, efficiency, and long-term opportunities, whereas employees face the practical consequences when new tools disrupt established workflows. Kumar argues that workers are more likely to adopt AI when they feel involved in the decision-making process, understand how it will benefit them, and trust that the technology is being implemented to support—rather than threaten—their work.
De Neve maintains that an “augmentation” approach creates a different relationship between workers and AI. “Opting for augmentation triggers a cycle of cumulative growth. Employees engage with curiosity and agency, becoming ‘pilots’ of the technology rather than mere passive passengers.”
The danger of prioritizing automation becomes particularly evident when companies cut back on hiring young talent or eliminate entry-level positions. These roles are often viewed as easy to automate because they involve repetitive tasks, basic analysis, or administrative work. Yet, they are also the environments where future leaders learn to exercise judgment, build relationships, and gain an understanding of customers. “When compensation and entry-level roles are cut to fund technology, the internal leadership pipeline is depleted, and the psychological safety needed for innovation is destroyed,” states De Neve. “The battle of the AI revolution will not be won by organizations that replace people the fastest, but by those that best empower them.”
The question is not whether companies should ignore efficiency, but whether it becomes their sole strategy. Professor Guillaume Coqueret of Emlyon Business School argues that AI adoption depends as much on individual adaptability as on organizational factors—such as leadership, incentives, culture, and the choice of tools. Traditional organizations face a greater challenge, as they often must integrate AI into legacy processes, ingrained habits, and varying levels of trust.
This alignment also impacts quality. An emerging risk is so-called “workslop”:
AI-generated content that appears useful but lacks the substance needed to move work forward. Alan Lerner, a professor at the Open Institute of Technology (OPIT), defines workslop as “AI-generated work content that looks like good work but lacks the substance needed to meaningfully advance a specific task.”
The danger lies in the fact that AI increases output volume while shifting the burden onto employees, who must review and interpret ambiguous results; this creates a false sense of productivity while simultaneously increasing the effort required to ensure AI outputs are reliable.
Data from Prosper Insights & Analytics points to similar concerns. Among surveyed employees, 37% state that AI requires human oversight, while 34% worry that it might provide incorrect information or generate “hallucinations.” These responses suggest that many employees do not reject AI; they understand that its value depends on human judgment.
Maintaining that judgment becomes more difficult when employees feel anxious, overwhelmed, or uncertain about how their work is evaluated. Debora Nozza, an assistant professor at Bocconi University, notes that AI can reduce the burden of certain tasks while simultaneously increasing the mental load associated with others.
“In theory, AI should reduce workload. In practice, it often changes the nature of that workload,” says Nozza.
Prosper Insights & Analytics also revealed that 17% of employees report feeling anxious about AI. This does not mean AI adoption should be halted, but companies must pay closer attention to morale, training, communication, and workload. If AI limits employees’ ability to exercise judgment, the anticipated efficiency gains could prove lower than leaders expected.
The most successful AI strategies will likely be evaluated using criteria that go beyond adoption rates or cost savings. Companies will need to consider whether AI is improving work quality, helping employees make better decisions, boosting talent retention, and developing the next generation of leaders.
Automation can streamline legacy processes. Augmentation can help companies develop new sources of value. Companies that recognize this will be better positioned to transform AI from a short-term efficiency tool into a lasting source of competitive advantage.
Note: The consumer sentiment study mentioned above was conducted by my company, Prosper Insights & Analytics. It utilizes the same dataset used by the National Retail Federation and available via Amazon Web Services, Bloomberg, and the London Stock Exchange Group for comparative economic analysis.
Jobs and the Future of Work
Why AI is replacing some jobs faster than others
This article is from the World Economic Forum’s “Jobs and the Future of Work” section and was written by Atul Kumar—a student leader, innovator, and changemaker working at the intersection of technology, sustainability, and youth empowerment. He is passionate about using artificial intelligence and systems thinking to solve global development challenges and frequently participates in discussions on educational equity, climate action, and emerging technologies.
Jobs in the financial sector, characterized by an abundance of data, are the most likely to be taken over by AI.
Our Impact
What is the World Economic Forum doing to accelerate action on jobs and the future of work?
Overview
Explore and track how artificial intelligence affects economies, industries, and global issues.
Stay up to date:
Artificial Intelligence
Follow
Data-rich industries are the most likely to undergo disruptive transformation due to AI.
Data-scarce industries are striving to digitize in order to reap the benefits of AI, yet they face greater challenges as they clash with established practices.
Employees and job seekers should focus on opportunities that combine technological capabilities with human judgment and business needs.
Everyone is talking about how AI will take over jobs, and that is partly true. We worry about AI taking over jobs, yet we do not fully understand how it will happen, when it will occur, or how much time we have to prepare.
Most people believe that task complexity determines whether AI adoption will be fast or slow. This assumption is completely wrong. As Sun Tzu’s maxim goes: “Know yourself and your enemy, and you will always be victorious.” Therefore, it is essential to understand the following dynamics. Clarifying the impact of AI on value creation, employment, and productivity
AI skills are becoming more important than work experience: here’s how job seekers can adapt
How can you harness the potential of generative AI to land a better job? Experts weigh in
How AI works
The inner workings of AI remain somewhat murky, even to experts. Yet, we know one crucial thing: it learns.
Have you ever received an electric shock while handling electronic devices?
It has happened to me, and that experience taught me not to do it again. Humans learn from experience.
AI models learn from data. An AI model with limited data is like a small child. In contrast, one with vast amounts of data is like a seasoned grandparent. The data paradox
What is harder: driving a car or writing code? Most people would say programming. Yet, in the realm of AI development, the opposite seems to be true.
Large Language Models (LLMs) are relatively new. Before ChatGPT, few people associated AI with chatbots; instead, they linked it to *Terminator*. The LLM era began around 2013–2014 with neural networks like word2vec. Autonomous driving, on the other hand, got its start in the 1980s. In 1987, Ernst Dickmanns’ team managed to get a Mercedes-Benz van to drive itself at 96 km/h on a German autobahn using computer vision.
Despite this massive head start, autonomous vehicles still lag behind LLMs. While ChatGPT performs reliably in countless situations, AI drivers continue to show inconsistent performance.
But why? Companies like Tesla and Waymo have invested billions. Yet, if a new company wanted to enter this sector—even with brilliant engineers and unlimited funding—it would still need thousands of hours of diverse driving data.
Some types of accidents are so rare that it is nearly impossible to train systems to handle them.
In contrast, LLMs are trained on the entire internet—a data-rich environment. Consequently, AI is more likely to replace programmers than drivers; not because programming is simpler, but because obtaining the necessary data for it is easier.
AI is essentially like that college student who had access to all the old exams and study guides. Naturally, they will ace the test compared to someone trying to get by with incomplete notes from just a few classes.
That is exactly what is happening in the job market. Some industries are swimming in useful data that AI can learn from. Others? They are working with scraps. The figures are quite revealing: industries with vast amounts of high-quality data could reach AI adoption rates of 60% to 70%. In contrast, data-poor sectors might struggle to surpass 25%.
The jobs most affected
Software development is experiencing a massive impact. GitHub hosts over 420 million repositories, including at least 28 million public ones—millions of examples of how to solve programming problems. Tools like GitHub Copilot analyze all that code and learn to write programs autonomously. Three out of four developers already use AI assistants.
Customer service is another highly vulnerable sector.
It is an ideal candidate for AI automation thanks to the abundance of data. IBM notes that AI uses information from calls, emails, and ticketing systems to improve responses and cut costs by 23.5%. The financial sector also makes extensive use of machine learning in algorithmic trading; high-frequency trading accounts for approximately 70% of U.S. stock market volume, leveraging vast amounts of market and transaction data to make predictions.
Data-scarce sectors
AI adoption in the healthcare sector is progressing slowly due to a scarcity of public data. Less than 10% of surgical datasets are publicly accessible, owing to HIPAA restrictions and data fragmentation. Patient data is scattered across various hospitals, insurance companies, and clinics. AI cannot learn effectively when information is siloed in countless different locations.
Construction may be the industry most resistant to AI—not because building houses is rocket science, but because the sector maintains very few digital records.
Every project is unique, documentation is often poor, and there is no standardized way to track what works and what does not.
The potential for AI in education is constrained by student privacy laws. The U.S. Department of Education notes that FERPA restricts data collection and sharing, thereby limiting AI’s use of student information.
Faced with this data scarcity, some industries are turning to intrusive measures. Hospitals are installing comprehensive video surveillance systems in operating rooms—ostensibly to train surgical AI systems—but this results in unprecedented surveillance of medical professionals. AI-based monitoring systems are being developed to track students’ eye movements, facial expressions, and typing patterns during exams, raising concerns about surveillance systems that could far exceed their original purpose.
The Economic Reality
The AI-driven transformation does not affect all economic sectors equally. In data-rich sectors, we are witnessing what economists call “creative destruction” at breakneck speed. Traditional jobs vanish almost overnight while new ones emerge; however, these new positions often require entirely different skill sets and tend to cluster in technology hubs. A customer service center that once employed 500 people might transform into a team of 50 AI-oversight specialists working from a single location.
Industries with limited data availability face a very different challenge. They must digitize to remain competitive, yet this process creates constant friction between cutting-edge technology and established practices. Transformation proceeds more slowly but has a deeper impact, restructuring entire departments rather than merely replacing individual jobs.
While new jobs are indeed emerging, it is not a simple one-for-one swap. Projections suggest that 92 million jobs will disappear by 2030, while 170 million new ones will be created.
However, these are not direct replacements occurring in the same locations or involving the same people. The true challenge lies not just in the number of jobs, but in the gap between where positions disappear and where they reappear, as well as the disparity between the skills workers possess and those required by new roles.
How to Adapt
For job seekers, the key is understanding that the boundaries between sectors are blurring faster than new job categories are solidifying. Instead of focusing solely on traditional career paths, look for roles that bridge different fields—positions that combine human judgment with AI capabilities, or that act as a link between technical systems and business needs.
Reassess your skill set by prioritizing adaptability over mere technical experience. Instead of simply listing your past actions, identify how you learn, solve problems, and work with new systems. Employers increasingly value individuals who can navigate uncertainty and integrate new tools into existing workflows. Your ability to successfully adopt significant workplace changes may be more relevant than your mastery of specific software.
Focus on friction points. Every organization implementing AI faces the same challenge: how to make sophisticated technology work within complex—and often chaotic—human systems. Look for roles involving management, training, or process optimization at companies adopting AI. These positions often do not require deep technical expertise, but rather individuals who understand how organizations actually function when theory meets practice.
Consider “last-mile” opportunities in your current sector. While tech hubs grab the headlines, every industry needs people capable of bridging the gap between AI capabilities and local implementation. Healthcare systems need professionals who understand both patient care and data analysis. Manufacturing plants require operators capable of working alongside automated systems. Often, combining your existing industry knowledge with a basic understanding of AI creates more opportunities than starting from scratch in an entirely new field.
Generative AI might not replace jobs, but it could replace skills—get ready
This article comes from Visier, a company that describes itself as revealing the human reality behind the data.
Keep your talent strategies agile in a fast-paced sector. Lead the way in attracting and retaining exceptional talent by gaining visibility into your hiring funnel, key success factors, and turnover risk.
About the author: Sarah P. Grant is a Montreal-based writer specializing in technology. She brings keen analysis and an approachable style to Visier, covering topics ranging from gender pay equity and Brexit to AI, burnout, and more. She enjoys spending time with her three children at the family sugar shack and creating abstract art.
To effectively upskill the workforce and make the right investments in learning, employers must anticipate future skill needs within the context of their business and the generative AI revolution.
Skills training and generative AI are not mutually exclusive
Popular topics
Culture | Leadership | Product and innovation | The future of work | Introduction to People Analytics
When IKEA began using an AI agent to handle frequent customer service inquiries in 2021, the move could have spelled serious trouble for the human workers displaced by the bot.
However, instead of mass layoffs, the company chose to retrain and reassign more than 8,500 call center employees to work as interior design advisors. This internal mobility initiative is part of a long-term strategy to boost sales among Gen Z shoppers.
If more employers adopt this approach—strategically prioritizing employability during periods of technological disruption—they could help mitigate worker anxiety regarding automation. In fact, experts are making bold predictions about the impact of generative AI across all industrial sectors, and the fear of the disruption it causes is widespread: according to a global employee survey conducted by Visier—involving over 3,000 participants from the US, the UK, and Germany—51% of respondents expressed genuine concern that their current skills could be replaced by AI.
While AI contributed to the loss of nearly 4,000 jobs in the US in May 2023 (based on data from Challenger, Gray & Christmas), the case of IKEA demonstrates that total automation is not inevitable.
Generative AI could replace up to half of the tasks in certain occupations.
If history is any guide, as more organizations leverage generative AI to support new business strategies, gain efficiency, and outpace the competition, some jobs will disappear while new ones are created. However, jobs requiring non-repetitive manual dexterity—such as those in construction—are likely to remain unaffected by automation for the foreseeable future.
Highly skilled office jobs will not disappear entirely, but they will undergo a radical transformation. As noted in this BBC article, humans will likely continue to perform roles that require strategic planning (such as business consulting), although even these roles are not immune to the impact of AI.
“In many cases, there is no immediate threat to jobs, but the tasks [emphasis added] will change,” Joanne Song McLaughlin, an associate professor of labor economics at the University of Buffalo, told the BBC.
A Goldman Sachs report published in March 2023 supports this claim: according to the document, AI is likely to transform two-thirds of jobs in the United States and Europe. These occupations involve tasks—ranging from 25% to 50% of the total workload—that could be replaced by AI. Similarly, a McKinsey report estimates that half of current work activities could be automated between 2030 and 2060.
To better understand the current state of skills training, Visier conducted a global survey of more than 3,000 employees across the United States, Canada, the United Kingdom, and Germany.
Successful organizations will know how to foster the best possible collaboration between AI and people.
To help their workforce adapt, leading companies are focusing on both *upskilling* (training employees for new roles) and *reskilling* (training employees to better perform their current roles).
Many people expect exactly this: according to a Visier survey, 86% of employees expect their company to play a role in reskilling to ensure they are not easily replaced by AI tools; in fact, 63% consider this to be the sole responsibility of the employer.
While some individuals can effectively manage their own skills development, relying solely on employees to drive reskilling initiatives is a risky strategy.
Employers who take a “wait-and-see” approach to skills development may lack the agility needed to compete in a workplace where AI is rapidly evolving. In fact, 77% of business and HR executives consider flexible skills deployment crucial for navigating future disruptions. Therefore, rather than viewing the relationship between humans and AI as an either-or choice, forward-looking employers are taking the initiative to analyze how people and machines can complement each other to achieve organizational goals.
3 actions organizations must take now to prepare their workforce for AI
To effectively upskill their workforce and make the right training investments, employers must forecast future skill needs within the context of their business. This involves anticipating which agent—human or AI—will handle specific tasks at particular times and staying abreast of the emerging capabilities of generative AI (GenAI).
- Project a skills training timeline
Rushing to adopt AI before its capabilities are mature enough to perform a specific function can have serious consequences. For instance, the tech news site CNET had to temporarily halt the publication of AI-generated content after being forced to issue corrections for 41 out of 77 articles produced using the technology.
On the other hand, leading companies cannot afford to wait too long to jump on the AI bandwagon, as they risk falling significantly behind the competition.
To further complicate the picture, the projected timelines for the rollout and development of new AI capabilities are constantly shifting: as McKinsey noted in its report on generative AI, the firm’s recent estimate—suggesting that half of current work activities could be automated between 2030 and 2060—is “roughly a decade earlier than our previous estimates.”
- Define skill requirements based on tasks humans can perform but AI cannot
Future skill requirements must also be specific enough for business leaders to know where to invest in training. In the past, many analysts pointed out that machines lack empathy and that, consequently, people would never find comfort in a robot. Yet, some doctors are already using ChatGPT to improve their patient interactions and deliver bad news.
This means that when developing the human skills needed in the AI era, organizations must take a deeper look at which specific elements of relationship-building can be automated and which cannot.
- Stay abreast of rapid AI advancements
Finally, given that generative AI is being deployed even as it is being refined, the initial skills required to interact with early versions of ChatGPT may soon become obsolete. For instance, crafting effective prompts (instructions for AI) might cease to be a key competency for those not involved in actual AI development. A new generation of tools will enable better results from less precise instructions.
With every new leap in AI capabilities, business leaders will face the same questions: Where do machines excel, and where do people excel? How can they collaborate to drive the organization forward? It is a puzzle that will require an ongoing solution.
Vee is Visier’s digital AI assistant. Discover how it can deliver instant workforce insights to anyone, anywhere. Planning ahead is possible, even when the future is uncertain.
Although the pace of change has recently been dizzying, it is worth remembering that many experts estimate there is only a 50% chance of achieving human-level AI by 2059. Therefore, at least for the foreseeable future, AI will not be able to outperform humans in many tasks.
As academic and business expert Oguz A. Acar states in this HBR article, an enduring human skill is “problem formulation”—the ability to “identify, analyze, and define problems.”
He argues that problem formulation “could prove as crucial as learning programming languages was in the early days of computing.” Yet, 85% of senior executives believe their organizations are poor at defining problems. It is therefore worth considering whether this is an area requiring investment in training.
Thus, while it is difficult to predict with absolute certainty how AI adoption will evolve in the long term, it is essential to understand which necessary skills cannot be automated in the near future.
It is also crucial to thoroughly assess existing competencies. By taking a strategic approach to skills data, organizations can gain a comprehensive view of their workforce’s current capabilities. Comparing current reality with future needs creates a clear picture of where the organization should focus its skills development efforts.
It is up to employers to identify who possesses the right skills and who needs training to acquire them. To navigate the future disruptions brought about by AI, organizations must foster a culture of talent mobility and continuously address skills gaps.
The benefits are clear: reskilling offers employees the opportunity to enhance their capabilities and remain valuable assets to the company. From the employer’s perspective, this also means that human skills are not valued merely as an afterthought—that is, long after a hasty decision to automate has already been made.
How workers discover new ways of working
The following contribution comes from OpenAI, which describes itself as follows: OpenAI is an artificial intelligence research and deployment company. Our mission is to ensure that artificial general intelligence benefits all of humanity.
Our vision for the future of AGI
Our mission is to ensure that artificial general intelligence (AI systems that are smarter than humans) benefits all of humanity.
Authored by the team.
New economic research from OpenAI shows how workers use AI to perform tasks outside their usual occupations and which ones they incorporate into their workflows.
Workers use AI differently when they go beyond their standard roles
Tasks from other occupations can become recurring parts of new AI-driven workflows
Some new activities become more established than others
From task crossing to job expansion
Our latest research highlights the importance of how tasks are divided and executed, alongside access to AI tools. These new tasks—outside workers’ usual roles—become part of their daily AI usage. If these activities evolve into regular responsibilities, job roles could expand in scope even while retaining the same title.
Our first report, *Work at the Frontier*, documented “task crossing”: workers using AI to perform activities traditionally associated with other occupations. Our latest research examines what happens next.
After analyzing over 1.5 million work-related ChatGPT messages between April and July 2026, we found evidence that some cross-occupational AI usage is recurring. Workers revisit tasks that fall outside their usual occupational boundaries, and over time, these activities account for a larger share of the AI usage we observe.
The findings offer an initial look at how access to AI can lead workers to move from simply trying out new tasks to integrating them into their regular routines. For companies seeking the best way to adopt and leverage AI, trends identified in our four-month study on professional AI and prompt usage offer insight into how to reduce friction between identifying a problem and making progress on the broader task with AI assistance. Our findings indicate that work design deserves as much attention as access to AI tools when implementing AI strategies within organizations.
Workers use AI differently when stepping outside their core roles
Workers utilize AI in distinct ways depending on whether a task aligns with their specific professional role. They craft different prompts for tasks that fall within their usual duties compared to those that lie outside them. When inquiring about tasks outside their occupation, they write shorter prompts on average than for tasks central to their role. They are also less likely to request explanations, step-by-step instructions, specific response formats, or advice.
At the same time, they are more likely to provide examples or context when asking AI for help with matters outside their occupation. They are also more likely to ask AI to check or verify information.
One possible interpretation is that workers use AI to tap into specialized knowledge from other domains. Rather than asking AI to teach them an entirely new field, workers may present a problem and provide relevant context—such as a document, an example, or information from a colleague—to help them apply knowledge from another field to their work.
Tasks from other occupations can be recurrently incorporated into new AI-driven workflows
The next question is whether workers revisit these activities. Among the approximately 6,200 workers observed continuously between April and July, tasks from other occupations that they had previously performed rose from 13.1% of AI activity linked to specific occupations in April to 25.9% in July. This pattern suggests that workers are incorporating assistance for tasks outside their usual roles into their regular workflows, rather than simply trying them out once.
Another analysis helps put this recurrence into context. In a sample of comparable observations tracked into the following month, workers revisited a task from another occupation—one they had used the previous month—23.6% of the time; by contrast, similar workers who had not been observed using that task the previous month used it only 8.4% of the time. Similar differences appear regarding tasks specific to the occupation and general tasks.
Some new activities become more established than others
Recurrence rates vary considerably depending on the type of work. Workers revisited certain tasks from other occupations the following month with relatively high frequency:
Discussing goods or services with customers: 54%
Drafting advertising or promotional copy: 44%
Creating marketing materials: 37%
By comparison, workers revisited the task of explaining financial information approximately 15% of the time. The average rate of repetition the following month for tasks from other occupations was 18.5%. These differences may reflect areas where AI naturally fits into recurring workflows. They could also reflect differences in workplace norms, levels of caution, or the perceived consequences of making a mistake.
From crossing into other tasks to expanding job roles
The first two reports in the *Work at the Frontier* series show how the nature of work can change before job titles do. The initial report showed that workers use AI to cross traditional occupational boundaries. This report concludes that some of these activities become part of routine workflows. These patterns point to a potential AI-driven pathway for job transformation: a worker tries an activity outside their traditional role, finds AI helpful for performing it, and begins to incorporate it into their regular work.
AI can transform jobs long before job titles change: workers use it for activities associated with other occupations and adopt some of them over time. If these activities become routine responsibilities, the scope of a job’s activities could expand even if the title remains the same. We will continue to study these changes to better understand how AI is transforming the division of labor and what this implies for workers, companies, and the economy as a whole.
Footnotes
We identified occupations based on information regarding job roles or departments provided during onboarding for ChatGPT Business. The messages come from a sample of these users’ conversations. We excluded data where training usage is disabled and messages that could not be usefully classified. We carefully protect user privacy by aggregating and anonymizing messages for analysis. During this work, researchers on our team never read individual user messages. All messages are aggregated and anonymized for analysis; we never read individual messages during this process. As workers try out new tasks in a given month, the number of tasks that can be considered recurring increases.
Why 42% of Companies Just Abandoned AI (and How Those That Didn’t Are Succeeding)
The following contribution comes from Medium, a platform that defines itself this way: Everyone has a story to tell.
Medium is a home for human stories and ideas. Here, anyone can share knowledge and wisdom with the world without needing to build a mailing list or an audience first. The internet is a noisy, chaotic environment; Medium is a quiet space, yet one filled with valuable ideas. It is a simple, engaging, and collaborative platform that helps you find the right readers for whatever you wish to express.
Ultimately, our goal is to deepen our collective understanding of the world through the power of writing.
Authored by Nayeem Islam, an expert in engineering solutions.
I saw a CTO on the verge of hurling his laptop across the conference room.
Not literally, but close.
His team had spent eight months and $400,000 developing an AI system designed to revolutionize their customer service. The demo? Incredible. The presentation to the board? They loved it. Then, they rolled it out to real customers.
The system failed within two hours.
“We did everything right,” he told me later. “The latest GPT models, top-tier engineers—we followed all the best practices. Why didn’t it work?”
He wasn’t the only one. Far from it.
A global survey has just revealed striking figures: 42% of companies abandoned the majority of their AI initiatives in 2025. That’s up from 17% in 2024. Almost half. And most notably: the average company scrapped 46% of its AI proofs of concept before they even reached the production phase.
But something made me stop and think.
I spent the last six months studying the companies in that other 58%—the ones that didn’t give up. The ones whose AI actually works. And I discovered something curious: they don’t use better models, nor do they spend more money.
They use a completely different architecture.
An architecture that has resided in your brain for millions of years.
In this article, I’ll show you why traditional enterprise systems kill AI projects, what brain-inspired design really means in practice, and what the MCP protocol is—a technology that has seen explosive growth, reaching over 36,000 implementations in just six weeks. You’ll see real code, concrete figures from companies that made this work, and why the solution has nothing to do with AI itself.
The real reason your AI initiative will likely fail
Let me set the scene.
You have a CRM that communicates with your HR platform via an API. Your financial software connects to three other systems using custom integrations. Contract management has its own database. Everything is interconnected in a massive hub-and-spoke structure: there are APIs everywhere and custom code holding it all together—like duct tape on a race car.
Then, someone in senior management says, “We need AI.”
So, you try to layer AI on top. The AI needs data from six different systems. Each has its own authentication system, data format, and update schedule. The AI can’t adapt because everything is hardcoded. A single API change is enough to bring the whole system crashing down.
Does this sound familiar?
Press Enter or click to view the image at full size.
I call this the “more AI” trap. You are forcing AI onto an architecture that was never designed for it.
It’s like putting wings on a car and wondering why it doesn’t fly.
The data confirms this emphatically. A RAND Corporation analysis reveals that over 80% of AI projects fail—double the failure rate of conventional technology projects. Another study concluded that 90% of AI initiatives never even reach business users; they die during the integration phase, suffocated by their own complexity.
Last quarter, I analyzed 23 failed AI projects across various companies. The pattern was always the same:
A magnificent proof of concept.
Excited stakeholders.
A massive budget.
Total failure when moving to production.
Why?
Because we are building AI systems exactly the same way we built traditional systems. Rigid. Static. Deterministic. A single path from A to B; if anything changes along the way, everything breaks.
But AI doesn’t work that way. AI needs flexibility. It needs context. It must adapt when things change—and in production, things are always changing.
Your brain figured this out 500 million years ago.
What your brain does with a billion bits per second (that we don’t)
Okay, let’s run a quick experiment.
Right now, as you read this, your brain is processing the words on the screen. But it is also managing the sensation of the chair, the sounds in the room, the temperature, perhaps hunger, your posture, the time, your to-do list…
According to Caltech research published in December 2024, your sensory systems gather approximately one billion bits of information per second from your environment.
But here is the surprising part. Your conscious mind—the part that actually makes decisions—processes about 10 bits per second.
Do the math. That means filtering out 99.999% of the information.
Your brain doesn’t just ignore most of the data; it ruthlessly discards almost all of it, keeping only what matters for what you are trying to do at that moment. And this isn’t a flaw; it’s an advantage. It’s what keeps you from going crazy.
When I first read about this research, it completely changed how I understood enterprise data architecture.
Most companies do exactly the opposite. They save everything. Every customer click, every API log, every transaction, every email. They dump it all into a “data lake,” hoping someone will strike gold there later.
Then they wonder why their AI models hallucinate or give terrible answers.
The AI isn’t broken; it’s drowning in noise. How your brain actually works (the parts relevant to AI)
Your brain isn’t just one big processing unit. It is made up of specialized regions that focus on specific tasks but coordinate seamlessly.
Your visual cortex processes images. Your auditory cortex handles sound. Your temporal lobe manages memory. Each part performs its function exceptionally well.
But here’s where the magic happens: your frontal lobe acts like an orchestra conductor. Neuroscientists call this “executive function.” It decides which brain regions to activate for each task. It doesn’t transmit data along rigid pathways; it coordinates activity dynamically based on what you are trying to achieve.
Need to remember where you left your keys? Your frontal lobe doesn’t just call a pre-programmed “memory API.” It activates spatial memory centers, visual memory, and even emotional memory—if, say, you were stressed when you put them down. The pathway varies depending on the context.
This is the architecture we need for AI systems.
People are largely accepting of AI taking on many jobs—but only up to a point
This article is from the Harvard Business School *Working Knowledge* portal.
Authored by Julie Jette, with contributions from James W. Riley.
Americans may be more open to AI taking over work tasks than previously thought. Research by James Riley reveals which jobs people are willing to see automated, where they draw the line, and why ignoring the value of the human touch could backfire for companies.
People are largely accepting of AI taking on many jobs—but only up to a point
Despite concerns that artificial intelligence might displace large numbers of workers, people say they are willing to let machines handle many tasks—and even entire occupations—especially if AI can do the job better, faster, and at a lower cost. Only a few professions are considered off-limits for automation, including the clergy and childcare workers.
Overall, the public supports automating approximately 30% of jobs based on current AI capabilities, according to a survey of 2,357 people covering 940 occupations. Furthermore, when participants were asked to imagine a more advanced future version of AI—capable of outperforming humans at a lower cost—support for automation nearly doubled, reaching 58% of occupations.
This sharp increase suggests that resistance to AI-driven job automation stems primarily from doubts about the technology’s technical feasibility, rather than from social or ethical objections to replacing human labor, says James Riley, an assistant professor at Harvard Business School and the survey’s lead researcher.
On the other hand, respondents expressed ambivalence or moral opposition regarding the automation of a significant number of jobs, even if AI could perform the work flawlessly and more cost-effectively. “Some employees will resist for their own reasons,” notes Riley. Even customers might be reluctant if companies “bypass their key audience—the people who value their products, services, and brand. Therefore, there is undoubtedly a risk involved in automating jobs or replacing human tasks with technology.” The question is whether, thanks to these tools, we can accomplish even more in our jobs than we could have in the past.
Before deciding where and how to implement AI to boost productivity, Riley suggests that companies consider the areas and extent to which the public values a “personal touch.” Regardless of whether AI can outperform humans, companies risk backlash from both employees and customers if they allow technology to take over certain tasks.
Riley presents the study’s findings in a working paper titled “Performance or Principle: Resistance to Artificial Intelligence in the US Labor Market,” published in October and co-authored with Simon Friis, a postdoctoral researcher at the Laboratory for Innovation Science at Harvard’s Digital Data Design Institute.
People support the use of AI as a performance tool
It might seem logical that as AI capabilities improve, market forces would drive widespread job automation. However, Riley and his co-author point out that history tells a different story, as economic incentives alone do not always determine the adoption of new technologies. The paper cites public resistance to genetically modified crops, nuclear energy, and embryonic stem cell research as examples of potentially beneficial technologies that were initially held back by social, moral, and ethical concerns.
When it comes to AI, people show a greater willingness to use the technology as a collaborative tool to assist humans rather than as a substitute for them:
Around 94% of respondents favor using current AI to enhance human work.
That figure rises to nearly 96% when the survey describes a more advanced version of AI that might emerge in the future.
This suggests that most people are comfortable with the idea of technology helping humans perform their jobs faster and more efficiently.
There are certain types of products or market sectors where people care just as much about the manufacturing process as they do about the product being more affordable.
“The fact that we can now use calculators to perform complex calculus or algebra means that, instead of wasting time trying to do what have become rudimentary calculations, we can focus our attention on more complex problems,” says Riley.
“So, the AI optimist believes we won’t suddenly be replaced in a way that leaves none of us with a purpose, a sense of meaning, or a job. It’s about whether, thanks to these tools, we can achieve even more in our work than we could have in the past.”
Which roles should continue to be performed by humans?
However, regarding the replacement of humans by AI, opinions remain divided on the automation of nearly 42% of professions. Respondents indicated that technical advances alone are not enough to win their support for automating many roles—ranging from nuclear reactor operators and sports medicine physicians to bakers. They feel a strong moral objection to automating a select few occupations—about 12%—including funeral directors, athletes, and artists. “The respondents suggest there is a social and moral boundary defining what we, as individuals and as a species, should continue to do,” notes Riley. “We will need to have a serious conversation, grounded in rigorous analysis, about where these limits lie and why.”
The respondents’ demographic characteristics—age, gender, ethnicity, and political affiliation—matched those of the U.S. adult population.
Leaders must find the ethical boundary
The survey did not ask participants whether they would be comfortable with the automation of their own professions; Riley notes that public opinion regarding which jobs are susceptible to AI automation could, naturally, shift over time.
“It is easy for respondents to state all this in hypothetical terms, but they might not feel quite as comfortable when it actually starts happening,” he notes. “Because there will be more at stake—unlike the current situation, where the issue is perceived as something distant or merely a thought experiment.”
For now, Riley recommends that leaders seeking to balance AI usage while maintaining stakeholder trust should:
Prioritize performance when evaluating AI opportunities
Companies planning their short-term AI strategies should focus on areas where customers do not raise significant moral objections and where the utility of AI is proven, Riley suggests. Consider AI-related norms in long-term planning
As companies expand their use of AI, planning must go beyond technical limitations and performance to encompass social and organizational factors, including talent retention and demand-related risks.
Be transparent about AI usage
As AI capabilities expand, companies may wish to extend its use to more areas; however, Riley points out that they must openly communicate to stakeholders exactly how and why they are implementing this technology.
Understand the risks of crossing certain boundaries
For some customers, a human presence will be a non-negotiable requirement for certain tasks and functions. Executives in some sectors need to be aware that automation could diminish the value of their products or services and trigger customer pushback.
“There are certain types of products or market sectors where people care just as much about the manufacturing process as they do about the product being made at a lower cost,” Riley states.
McKinsey explains why AI won’t take your job, even though it can already automate 57% of all work hours in the U.S.
The following article is from *Fortune* magazine and was written by Nick Lichtenberg, Business Editor.
A new report from the McKinsey Global Institute addresses one of the modern economy’s most pressing fears: the mass job displacement that artificial intelligence threatens to cause. While McKinsey’s research indicates that current technologies could, in theory, automate about 57% of work hours in the U.S., the consultancy concludes that this high figure measures technical potential regarding tasks, not inevitable job losses.
Rather than mass replacement, research conducted by Lareina Yee, Anu Madgavkar, Sven Smit, Alexis Krivkovich, Michael Chui, María Jesús Ramírez, and Diego Castresana argues that the future of work will be defined by collaborations between humans, agents, and robots—all powered by AI. Their report, titled “Agents, Robots, and Us: Skill Collaborations in the Age of AI,” highlights that capturing the enormous potential economic value of AI—estimated at $2.9 trillion in the U.S. by 2030—depends entirely on human guidance and organizational redesign.
The Enduring Value of Human Skills
The primary reason AI will not immediately displace half the workforce is the enduring relevance of human skills. Although they will be applied differently, McKinsey’s analysis shows significant overlap in required capabilities: more than 70% of the skills employers seek today are used in both automatable and non-automatable jobs. This suggests that as technology adoption advances, most skills will remain relevant, though the ways and contexts in which they are used will evolve. For instance, highly specialized yet automatable cognitive skills—such as routine accounting processes and the use of specific programming languages—could face the greatest disruption. However, even as AI takes over tasks like document preparation and basic research, workers will still need to apply their existing skills in new contexts, focusing instead on formulating questions and interpreting results. A key point is that skills rooted in social and emotional intelligence—such as interpersonal conflict resolution, design thinking, negotiation, and coaching—will remain uniquely human, as they require empathy, creativity, and contextual understanding that are difficult for machines to replicate. Likewise, competencies related to care and assistance are likely to see the least change.
Redesigning work, not just automating tasks
For organizations to successfully leverage AI, transformation must go beyond automating isolated tasks within legacy structures. The report emphasizes that realizing projected economic benefits requires a complete redesign of workflows—revising processes, roles, culture, and metrics—to enable effective collaboration between people, agents, and robots.
Even in roles with high potential for technical automation, humans will remain essential for effective operation and for performing tasks that machines cannot handle. People provide vital oversight, quality control, and the indispensable human presence often preferred by clients, students, and patients.
This transformation is already driving massive shifts in the demand for new capabilities. Demand for AI proficiency—the ability to use and manage artificial intelligence tools—has surged sevenfold in two years, making it the fastest-growing skill in U.S. job listings. This competency, centered on collaborating with and guiding AI systems, demonstrates that the economy is rapidly adapting to new forms of collaboration.
Ultimately, while some individual tasks could theoretically be automated, the impact on employment will depend on whether organizations and institutions prepare people for the future. If history is any guide, employment is likely to evolve rather than contract.
The AI era is not about completely replacing the human workforce, but rather shifting the focus of human intelligence: moving from execution to orchestration and judgment. Just as the invention of the calculator did not eliminate mathematicians but instead freed them to solve more complex problems, AI automates routine tasks, allowing human workers to focus on complexity, decision-making, and caring for others. Ultimately, the work of the future will be a collaboration between humans and machines.
“AI integration will not be a simple technological implementation, but a rethinking of work itself,” the report noted. “It involves redesigning processes, roles, skills, culture, and metrics so that people, agents, and robots create more value together.”
For this article, Fortune used generative AI to help create an initial draft. An editor verified the accuracy of the information prior to publication.
Fortune Daily breaks down the traditional barrier between the audience and the newsroom. The program transforms Fortune’s trusted journalism into practical, accessible, and entertaining information for a new generation of business leaders. Watch it here.
Nick Lichtenberg
Nick Lichtenberg is a business editor at Fortune, responsible for reporting and editing a mix of breaking news and digital features. This includes stories developed with the help of artificial intelligence—published under the Fortune Intelligence section—as well as opinion pieces and book excerpts. Previously, Nick served as Executive Editor of Global News at Fortune. Before joining Fortune, he was a deputy editor at Business Insider—covering the economy, wealth, and real estate—and a breaking news editor at Bloomberg, where he covered corporate news and equity markets.
10 Reasons Why Automation and AI Cannot Replace Humans
The following contribution comes from Salient Process, a company that describes itself as follows: We take pride in being a fully remote company with a multinational presence. We offer IBM software in the United States, Canada, and the United Kingdom, with additional services in India and Colombia.
It is authored by Brian French, the company’s founder and CEO.
Is the future of the workplace destined to be led by artificial intelligence (AI) and executed by automated systems? Will AI replace humans? The short answer is no. The long answer requires a deeper analysis and an understanding of what AI and automation actually are. First, let’s break down these two technologies.
Table of Contents (Hide)
What are artificial intelligence and automation?
- AI has no physical form
- AI lacks emotional intelligence
- AI cannot match human creativity
- AI quality depends on its dataset
- AI lacks interpersonal skills (soft skills)
- AI still requires data verification
- Automation needs humans
- AI is inflexible
- Automation and humans have different limitations
- AI should complement humans, not compete with us
What are artificial intelligence and automation?
AI is a science that helps machines learn how to perform tasks and solve problems. From executing routine, repetitive administrative tasks to generating valuable, actionable insights from vast amounts of data, AI can complete many tasks that humans cannot—or tasks that would take humans much longer to finish.
There are many forms and applications of artificial intelligence. One of the most common and basic forms of AI is the language-model-based chatbot, where users can enter queries and receive results. For example, a user might ask an AI chatbot to draft a resume based on a specific set of skills; AI would use that data and its knowledge of information available on the Internet to draft the resume in question. These tools have limitations, which we will explore later.
Similarly, automation technology streamlines processes so that they require little to no human interaction. Automated systems improve efficiency and can perform tasks without interruption. For example, payment apps and automatic car wash systems use automated mechanisms to complete tasks that humans could also perform, but in a more efficient manner. In the workplace, robotic process automation (RPA) can handle simple, tedious tasks such as moving digital files, extracting data, and entering information.
With these terms and definitions in mind, we can analyze their limitations and discover how they can assist humans without replacing them.
- AI lacks a physical form
Some tasks require dexterity and physical intervention. With the exception of robots—which are limited to performing very specific tasks—AI and automation software lack the physical form needed to execute work outside the digital realm. For instance, AI cannot perform tasks requiring hand-eye coordination, nor is it suited for jobs that demand rapid changes in direction. While automated machinery can assist with labor-intensive actions like moving boxes, its development has not yet reached the point where it can fully replace human workers and their physical capabilities.
Some of the most common areas for AI implementation are warehouse planning and inventory management. AI can help design efficient layouts and management systems, but executing those plans is a task only humans can perform. Agriculture is another key sector where AI can provide assistance, yet human labor remains indispensable. Ultimately, AI is capable of performing only certain work functions; therefore, human presence is required to handle the more complex physical tasks that may arise in any workplace.
AI Lacks Emotional Intelligence
- AI Lacks Emotional Intelligence
Emotional intelligence (or emotional quotient, EQ) is a valuable asset in the business world. EQ is the ability to anticipate a reaction and act accordingly. In theory, the higher a worker’s EQ, the better they perform their job and the more effectively they interact with others. AI completely lacks EQ; in fact, working with a computer can sometimes be more intimidating than working with a person capable of feeling. Many roles require EQ, such as team leaders, teachers, counselors, and therapists.
Most AI systems attempt to mimic human intelligence but have no way to simulate or interpret emotions. To understand emotions, AI needs to be able to interpret human experiences, not just intelligence. Feeling emotions is what makes us human, and transferring this to a machine is likely impossible.
Beyond simply designing AI capable of understanding emotions, the system must demonstrate EQ when interacting with users. An apt way to describe this lack of EQ is that while you might trust what the AI tells you to do, you may not trust the way it communicates that information. Our emotional intelligence helps us determine how to approach problems in relation to ourselves. AI lacks a sense of “self” and struggles to guide us beyond simple, logical instructions. Method, motivation, and improvisation remain unique to human beings.
Jobs that do not require EQ or emotional connections are where AI and automation excel, although such opportunities are often limited. By using these tools to handle routine, repetitive tasks, we can better leverage EQ, giving humans more time to work on situations that actually require it. As previously mentioned, Robotic Process Automation (RPA) can be highly beneficial for performing simple, redundant tasks, allowing human workers to focus on more complex, creative work that demands emotional intelligence. 3. AI cannot match human creativity
Situations where AI assists humans can lead to a superior product. Thanks to AI, creative work is more accessible than ever. Using AI as a tool can help those with a creative vision achieve the results they seek.
However, allowing AI to take full control of creative projects has its limits. As humans, our creative vision constantly evolves, and we can find inspiration in situations and experiences at any moment. An AI-based creative tool generates an instant final result, leaving no room to develop or expand the concept beyond its initial stages. AI also operates based on pre-set data and does not deviate from it, whereas humans possess—by comparison—limitless creative potential.
AI and humans can achieve synergy in creative fields when collaborating toward a common goal. Yet, the current industry trend is to keep them separate or pit them against one another. Human creativity and emotional intelligence (EI), combined with AI’s efficiency and precision, can lead to significant achievements in creative work. While it is impossible for automation to reach the level of human creativity, using AI for specific aspects of creative work could prove beneficial.
- AI quality depends on its dataset
It is easy to mistake AI intelligence for genuine intelligence. AI software works by drawing on data to craft the response it deems most appropriate for your queries. The AI tool you choose can only perform tasks permitted by its input data. In other words, the data must be capable of guiding the AI to provide the answers and insights you need.
By providing the AI tool with more information—as well as up-to-date, relevant customer data—it can generate more accurate and useful conclusions. Consequently, human intervention will likely always be required to operate specific AI tools: humans supply data to these applications, and without data, AI cannot function. Ultimately, the effectiveness of AI depends on the data that human workers supply to it or generate for it.
- AI lacks interpersonal skills (soft skills)
Interpersonal skills—or “soft skills”—are non-technical competencies that define how people interact and work with others. These skills are highly valued in most companies across all sectors. Valuable interpersonal skills include respect, negotiation ability, confidence, friendliness, adaptability, the ability to foster teamwork, innovation, and creativity. AI lacks the interpersonal skills necessary to integrate effectively into a work team.
People develop interpersonal skills (or “soft skills”) through training and experience. These skills improve over time, and everyone—regardless of their job—likely needs to learn at least some of them. Machines cannot acquire these skills, just as they are incapable of understanding or feeling emotions. No AI or automation software can fully replicate the intelligent capabilities possessed by a human being; therefore, people will always be needed to integrate and lead teams in this way.
- AI still requires data verification
Not everything generated by an AI tool is perfect. Artificial intelligence creates a great deal of erroneous information that must be verified. AI analyzes available data to craft a natural-sounding response based on the language models it employs. This process facilitates the generation of false yet credible information on a large scale. Although using AI to monitor AI might seem like a good idea, this approach entails certain problems.
AI that relies on news sources and other websites to generate responses is vulnerable to misinformation. If it uses data containing factual errors, it will also produce false results. Furthermore, overlooking nuances or lacking specificity are common errors in AI-generated text. Consequently, human intervention remains necessary to verify results and make appropriate adjustments.
- Automation needs humans
Humans created AI and automation software, but the link between the two runs deeper. The strongest argument for why AI cannot replace humans is that automation still requires human intervention. The use of “human-in-the-loop” systems keeps us involved in a way that ensures the best results, whether in data verification or AI maintenance.
Whether it is a robot moving boxes in a warehouse or an RPA tool completing simple digital processes, humans are needed to oversee and correct these actions when necessary. In most cases, people are also required to train the automation tools and processes. Ultimately, it seems that human beings will always be part of the process when it comes to the use of AI and automation.
- AI is inflexible
Automation and AI software are designed to follow strict rules to generate the correct results. While these programs excel in certain fields, they can only operate within the confines of their rules. In contrast, humans can complete a task even when facing obstacles. This ability to improvise and use personal initiative to achieve a goal is where humans hold a significant advantage over machines.
For example, if an unforeseen problem arises in the workplace, it is far easier to instruct a team of human workers on how to shift strategies and solve the issue than to attempt to redirect automated machinery on-site. Ultimately, workplace automation is capable of performing only specific functions. Humans, on the other hand, are often well-equipped to handle diverse tasks and adjust their approach as needed.
- Automation and humans have different limitations
As humans, we can draw upon our emotions and experiences to shape the way we work. Furthermore, certain limitations are inherent to human nature: we get tired, need to eat, require motivation, and need to maintain good physical and mental health. Despite this, we possess many strengths that make us valuable. We are capable of learning, guided by common sense, and able to interact and communicate in order to collaborate and achieve great things.
While AI may possess limitless energy, it is bound by the scope of its code and data. Humans are more versatile, capable of performing everything from physical labor to tasks involving complex reasoning and research.
Moreover, AI lacks a moral compass or inherent boundaries. A case in point is a 2003 thought experiment in which an AI was instructed to manufacture paperclips. Ultimately, the AI’s calculations led it to convert the entire planet into a paperclip production facility. This lack of boundaries and guidance demonstrates that AI prioritizes its assigned task above all else. Even decades later, this example highlights a major weakness of AI—one that makes humans superior leaders and agents of action. 10. AI should complement humans, not compete with us
Broadly speaking, the purpose of artificial intelligence is to assist humans, not replace them. While automation is ideal for tasks involving repetitive movements or other monotonous work, AI is not a suitable substitute for jobs requiring reasoning, interpersonal skills, and adaptability. The focus should be on how AI can help us, rather than replace us or become something we rely on entirely.
As job requirements evolve, the need for human workers grows. In 2020, the World Economic Forum published a report indicating that AI would generate 97 million new jobs, demonstrating that there are many roles AI cannot replace. Clearly, new opportunities will emerge as AI advances. Humans and AI can complement each other without the need for competition.
Both humans and machines can outperform one another in various areas. Understanding the limitations of AI and automation allows you to make the most of their respective strengths. With the right automation system, your business can thrive, and your employees can achieve more than ever before.
Salient Process provides the tools you need to take your business to the next level. Through our Digital Business Automation solutions, you can optimize workflows, automate tasks, manage content, and perform many other key functions—all while empowering your workforce. Contact us today to learn more about what we can offer.