Oct 28, 2026 · 1h 8m
Listen from 25:09
Listen at 25:09
AI’s biggest gains may require rebuilding the workplace
This AI Expert Argues We're Underestimating AI's Productivity Impact — With Ethan Mollick
The episode argues that AI’s economic impact depends less on model capability than on whether organizations redesign work, incentives, and education around it.
- 1Companies are already finding AI returns, but isolated productivity gains remain difficult to scale across organizations.
- 2Reasoning models, tool-using harnesses, and personal agents could expand AI’s reach while intensifying privacy and security risks.
- 3AI will reshape jobs and learning unevenly, rewarding organizations that avoid locking themselves into today’s limitations.
Don't miss
Mollick’s thought experiment about organizations fully exploiting existing AI capabilities predicts major disruption to jobs even if AI progress stopped today.
The brief
Ethan Mollick argues that AI’s productivity gains are real but hard to see because individual improvements do not automatically become organizational or economic value.
The path to larger returns runs through leadership, experimentation, and structural redesign: cross-functional teams and broader builder roles could dissolve boundaries between coding, design, product, and strategy.
Mollick’s capability-overhang argument is the episode’s central tension: current systems are already underused, while companies keep waiting for better products instead of imagining new forms of work.
Reasoning systems, tool-using harnesses, and coordinated agents could push AI into mathematics, science, and operations, but personal agents also raise serious privacy, cybersecurity, and prompt-injection risks.
The conversation extends beyond software to factories, schools, coding, scholarship, and culture, where AI may invalidate some roles while increasing demand in areas that continue to grow.
The closing warning is strategic: organizations should build flexible architectures that can absorb future model improvements rather than optimize around today’s AI limitations.
What was said on this episode
43 statements · 26 positive · 10 negative · 3 mixed · 4 neutral
Organizations have not yet adapted sufficiently to AI.
“We haven't adapted organizations to AI yet.”
Listen at 3:55
Current AI primarily improves individual performance rather than coordinated organizational performance.
“AI is a very single-player experience right now where you get individual performance, but organizations are about many people performing at once.”
Listen at 4:47
Nearly all large companies are currently seeing some return on AI investment.
“There is almost nobody at this point in the data from my colleagues at Wharton back this up who isn't seeing some ROI from AI.”
Listen at 7:22
Mollick has not heard a large organization say AI is not paying off.
“I haven't heard of a single organization saying that AI isn't paying off at a large scale right now.”
Listen at 8:19
Companies are not broadly reducing AI use because they consider it useless.
“I don't see any sign of people saying they're pulling back because AI turned out to be useless.”
Listen at 9:54
Companies Mollick speaks with are receiving large returns from AI.
“companies are getting large returns that we're talking to.”
Listen at 10:03
Some executives use AI as an analyst to identify previously hidden company problems.
“C-level teams are using this as a super analyst to find problems inside the company they weren't able to identify.”
Listen at 12:32
Companies need failed AI experiments to learn how to transform.
“I think they have to be doing experiments that fail.”
Listen at 14:26
AI is blurring traditional boundaries among coding, design, and product roles.
“coders are doing design work, designers are doing coding work, product managers are doing everything.”
Listen at 16:01
Skepticism about AI's ability to create transformation is currently unwarranted.
“I don't think skepticism about AI's ability to create transformation is warranted at this point.”
Listen at 18:10
Manufacturing and other large organizations are undergoing substantial AI-related change.
“there is a huge amount of change in these organizations.”
Listen at 20:11
AI-driven business transformation is faster than prior digitization, cloud, and e-commerce shifts.
“Is this faster than digitization, cloud, e-commerce, any of the other things that have transformed American business the last 20, 30 years? Certainly.”
Listen at 22:39
AI will not automatically produce human flourishing.
“I don't think human flourishing is a given outcome from this.”
Listen at 23:27
AI can promote flourishing by reducing drudgery and improving science, education, healthcare, and advice access.
“the way you get human flourishing, right, is we free people from drudgery. We increase the rate of scientific advancement. We increase the rate at which people, everyday people get access to better education, better health care, better financial advice.”
Listen at 23:54
Existing AI capabilities could transform much work over the next decade without better models.
“we don't need better models for complete transformation of a lot of how we work in the next 10 years”
Listen at 24:44
AI-driven organizational transformation will occur.
“the transformation will happen”
Listen at 25:09
Organizations will exploit existing AI capabilities and transform within five to ten years.
“over the next five or 10 years, we will fill in the capability overhang. Organizations will transform.”
Listen at 28:34
AI transformation will occur regardless of further model development.
“we will see transformation happen one way or another”
Listen at 29:03
Leading coders increasingly manage AI-assisted coding rather than write code directly.
“The best coders I know are not writing code anymore.”
Listen at 29:40
Without practical robots, everyday urban environments will change relatively little.
“unless we get robots doing work for us, you'll still walk down a city street.”
Listen at 31:11
Technological revolutions produce disruptive social and economic change.
“technological revolutions result in disruptive change.”
Listen at 31:31
AI will cause disruptive change, though its distribution remains unpredictable.
“I can promise disruptive change”
Listen at 31:50
Mollick no longer considers AI slowdowns a likely possibility.
“I no longer think they are.”
Listen at 34:26
Mollick cannot identify a specific technical barrier likely to halt AI development.
“I don't have a technical point anymore where I can point to this and say, this is the big problem that's going to hold back AI development.”
Listen at 37:01
AI capabilities will continue making steady gains despite physical constraints.
“I don't see any stoppage to steady gains in AI ability.”
Listen at 37:39
AI progress will be constrained by available compute and power.
“It will be constrained by the amount of compute and power we have.”
Listen at 37:53
Reasoning systems improve answer quality, especially for mathematics and science.
“that creates much more high-quality answers, especially in areas like math or science”
Listen at 39:20
AI utility has surged through the combination of models, applications, and tool-using harnesses.
“harness plus model plus app, working together have been what led to the explosion of utility recently.”
Listen at 40:49
Personal agents require access to a recent frontier model.
“if you don't have a frontier model, you can't do this kind of work”
Listen at 42:54
Personal agents could expose or transmit users' medical records without their knowledge.
“could it decide to send your medical records somewhere you don't know? Yes.”
Listen at 44:42
Connecting AI agents to private data creates privacy and security risk.
“There is risk in that.”
Listen at 45:12
AI companionship has not yet produced the expected mass social effects.
“AI companionship didn't like, and everyone's saying, oh, I'd have a friend who's addicted to AI. But like, we haven't seen the sort of mass effect I think we were expecting right away.”
Listen at 48:05
Recent AI advances may mark the beginning of rapid robotics progress.
“this might be the moment where things start to take off”
Listen at 50:11
Using AI to bypass learning reduces learning.
“if you use AI to short circuit doing work, i.e. cheating but even accidentally cheating yourself by saying help me explain this problem and the ai answers it for you you don't learn”
Listen at 54:08
Using AI as a tutor improves learning outcomes.
“if you use AI as a tutor you learn”
Listen at 54:17
AI now provides a potentially universal high-quality tutor.
“We finally have a universal tutor of high quality in AI.”
Listen at 55:46
AI is accelerating scientific work while destabilizing established scientific practices.
“we're seeing acceleration of science at the same time as we're seeing the foundations of science come apart”
Listen at 58:09
AI will eliminate some job categories, transform others, and expand demand for others.
“there are job categories that get invalidated in ways we don't expect. And some jobs will transform, and some jobs will grow”
Listen at 1:01:34
Falling token costs will let AI perform more work at lower cost.
“that lower cost means AI does more stuff for less money in a very short period of time”
Listen at 1:03:34
AI token economics are creating genuine downward pressure on prices.
“there's a downward pressure on prices that is real.”
Listen at 1:03:54
Companies are unprepared for large-scale AI-driven customer requests.
“Companies are not ready for that.”
Listen at 1:04:57
Personal agents will eventually move users' money into higher-interest savings accounts.
“your dot eventually will be able to take the money and just move it into a high interest savings account.”
Listen at 1:05:19
Companies should exploit current AI while continuing to explore future technologies.
“you have to be able to both do exploitation of current technologies while still exploring for the future.”
Listen at 1:06:30
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Books & mentions
Some links are affiliate links — PodLume may earn a commission if you buy.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.