AI and the job landscape in 2026

Editor’s note: This article was adapted from the January 2026 issue of MIT Sloan’s monthly AI at Work newsletter.

With 2026 shaping up to be another big year for artificial intelligence, some MIT faculty and researchers recently shared what they’re focused on when it comes to AI and jobs.

Here’s what they monitor:

Difference in accuracy between humans and LLM

MIT Sloan, Professor of AI/Machine Learning Practice

“I’m going to pay attention to the difference in accuracy between humans and large-scale language models.

“Automating knowledge work using LLM is the main focus of many companies’ generative AI pilots. For certain tasks, LLM may not be accurate enough, and it may be tempting to conclude that the task is not suitable for automation using LLM. However, rather than comparing the LLM’s accuracy to the highest possible (i.e. 100%), it is better to compare it to the accuracy of the humans currently working and track changes. Accuracy gap between humans and LLM for that task. Perhaps humans achieve 95% accuracy, while LLM only achieves 90%.

“The important thing to remember is that as Frontier LLMs become more capable, their accuracy will continue to improve, while human accuracy will likely remain the same.Therefore, there is a good chance that LLM accuracy will exceed human accuracy for many enterprise tasks by 2026.

“What are these tasks? How much business value do they represent? How many jobs are at risk? These are some of the questions I’m looking at as we head into 2026.”

AI guardrails

Principal Researcher, MIT Information Systems Research Center

“My colleague I am passionate about research into the guardrails companies need to establish to effectively and safely deploy AI solutions without compromising compliance, values, ethics, and innovation. This is a difficult balance. Old governance strategies won’t work for AI due to the pace of change and other reasons. We will continue to focus on new practices that help organizations adapt their governance so that AI solutions can scale and sustain themselves over time. ”

What happens when humans outsource their creativity to AI?

MIT Sloan Professor of Applied Economics

“Plasticity is the brain’s ability to change its structure and function throughout life in response to experience. So when we stop solving differential equations, we forget how to do it. When we stop doing calculus, we forget how to do it. And when we stop using our brains to remember phone numbers and directions, we forget them. When the phone replaces our sense of direction, we forget and become dependent on Google Maps.”

“I think it’s okay to forget about direction. But what happens if we start replacing experimentation, creation, or what-if thought processes with AI? Do we really want to forget these activities? What about entrepreneurship? What about art? What about music? I think the creativity that humans have shown for centuries, real creativity, is infinitely better than what an AI entity can do. may try more things, with higher variance, and produce results that are worse than those that individuals build on each other.

“So how to implement AI is a primary concern. I wrote a new paper with Isabella Loaiza on this very topic, and I believe we need to think more deeply about it.”

Understand the inner workings of AI models

MIT Sloan Senior Lecturer in Management Communication

“Looking ahead to 2026, I’m focused on the study of machine interpretability to see what it tells us about the inner workings of AI models, and potentially increase their safety and integrity.

“As Anthropic’s Chris Olah says, [generative AI] The model is effectively grew through training Rather than being explicitly built or programmed, it leads to technologies that are unusually opaque.

Mechinterp aims to reveal how a particular neural network works and what actually leads to its output. When it comes to leveraging AI, we want to see how it can help users make better-informed decisions, and we’re also excited about the impact AI will have on broader society. ”

Scaling your AI solution

MIT Sloan, Senior Lecturer in Information Technology

“This will be the year that companies move from experimenting with generative AI and agents to finding viable solutions that create real value at scale.

“With all the hype around generative AI and agents, it’s important to focus on the right questions: What problem are we trying to solve? To get the answers, we need to find the right combination of AI, traditional IT, and human technology for each task in our solution.”

LLM data

haranjuDigital Fellow at the MIT Digital Economy Initiative and Assistant Professor at Johns Hopkins University

“I predict LLMization of data will be a major trend that will unfold in 2026 and beyond. By “LLMization,” I mean that internal enterprise data sources and private company databases (e.g., Apple Notes) will be easily accessible to LLM-based agents, rather than being accessible only to humans through existing user interfaces. ”

Rama Ramakrishnan He is a practicing professor of AI/machine learning at MIT Sloan. His interests focus on real-world business applications of predictive and generative AI techniques and the creation of intelligent products and services.

barbara wixom I am the chief researcher. MIT Information Systems Research Center. Her research investigates how organizations create business value from their data assets.

Roberto Rigobon He is a professor of applied economics at MIT Sloan and co-founder of the MIT Sloan Sustainability Initiative. collective confusion projectConsider improving ESG measures.

melissa webster He is a senior lecturer in management communication at MIT Sloan. She investigates the adoption and impact of ChatGPT and other generative AI in both professional and educational fields.

George Westerman He is a senior lecturer in information technology at MIT Sloan. Global Opportunity Forum. His research focuses on digital transformation, bridging the fields of executive leadership and technology strategy.

haranju He is a Digital Fellow at the MIT Digital Economy Initiative and an Assistant Professor at Johns Hopkins University. He studies AI agents and how they help people do their jobs.

Source link