TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Wharton professor Ethan Mollick’s new book, Co-Existence: The Next Phase of AI, argues that people can prepare for increasingly capable AI by using it, learning its limits and deciding where it belongs in their work and lives. The approach does not resolve fears about job losses or other harms, and the scale and pace of AI’s effects remain uncertain.
Wharton professor Ethan Mollick argues in his new book, Co-Existence: The Next Phase of AI, that people should learn to use AI and decide how it fits into their lives as the technology takes on more work. His advice comes amid public concern about job losses and AI’s wider effects, but it is a personal strategy for adapting, not a solution to the risks or a guarantee that workers will benefit.
Mollick, an associate professor at the University of Pennsylvania’s Wharton School and faculty director of its Generative AI Labs, says the shift from chatbots to AI agents changes how people interact with the technology. Agents can be given tasks to carry out over longer periods, rather than requiring a person to direct every step. He says their errors and need for constant guidance have become less of a barrier, though the source does not provide independent performance data to quantify that change.
In the interview, Mollick describes reports from coders who say they are much more productive with AI. He also gives a personal example: after asking an AI system to cancel an airline Wi-Fi subscription, he says it contacted customer service and the subscription was ended. That account illustrates a possible use, but it is not evidence that agents can reliably handle similar tasks for all users.
Mollick’s broader argument is that direct experience helps people understand what AI can and cannot do. He expects personal agents to make some everyday tasks easier by lowering the effort required to complete them. At the same time, he cautions against handing over skills without reflection: he worries that if he loses too much of his own ability, he may no longer understand how computers work.
A practical guide to living with AI
Humans Can Coexist With AI. Here’s How.
Use it. Learn its limits. Choose where it belongs. Wharton professor Ethan Mollick’s book argues that firsthand experience can help people adapt as AI takes on more work. It does not settle fears about jobs or guarantee that workers will benefit.
01 / The shift
From chatbots to agents
AI agents can be assigned a task and work through multiple steps. Their performance and need for supervision vary; the interview offers examples, not independent reliability data.
Then · Chatbot
Prompt, response, repeat
A person directs each turn of a back-and-forth exchange and decides what to ask next.
Now · Agent
Delegate a sequence
An agent may carry out several steps toward a goal, such as contacting customer service.
Keep in view
People stay responsible
Check important outcomes, protect sensitive information, and keep skills needed to judge the result.
02 / Why it matters
Prepare without overclaiming
AI could change how work is organized and make some everyday chores easier. The consequences depend on reliability, adoption, human oversight, and how gains are shared.
71%
of Americans expected AI to lead to fewer jobs over the next two decades, according to a Pew Research Center study cited in the source article.
The result was seven percentage points higher than two years earlier. It records what respondents expect; it is not a forecast, proof of job losses, or a count of jobs already lost. Individual familiarity cannot by itself prevent displacement or determine who benefits.
03 / A way to learn
Try, check, decide
Mollick recommends practical experimentation to build a clearer sense of what AI can and cannot do, while retaining skills that matter.
Choose a task
Start with something low stakes and familiar.
Try the tool
Give it a concrete goal and observe its steps.
Verify the result
Check accuracy, security, and what needed your judgment.
Keep useful skills
Decide where delegation helps without giving up understanding.
04 / Examples & limits
Promising stories are not proof
Mollick shares reports of productivity gains and a personal example. The source does not establish how typical or repeatable these outcomes are.
Anecdote · Everyday task
Canceling Wi-Fi
Mollick says an AI system contacted airline customer service and ended his Wi-Fi subscription. It illustrates one possible use; it does not show that agents can reliably do this for everyone.
Caution · Skill retention
“If I de-skill myself entirely…”
Mollick worries that losing too much of his own ability could leave him unable to understand how computers work. Delegation should leave room for learning and human judgment.
Reliability
How often do agents complete different tasks correctly?
Oversight
How much guidance is needed, and who remains accountable?
Work & gains
Which jobs change, how quickly, and who receives the benefits?
05 / What remains open
Questions that experience cannot settle
Learning to use AI can inform personal choices. It does not answer broader questions about employment, safety, security, or how society shares the effects.
The survey measures expectations about the future, not what will happen to employment.
The interview does not quantify success rates or supervision needs across settings.
How quickly agents become common at work and at home remains uncertain.
Familiarity alone cannot determine who benefits or prevent flawed deployment.
06 / Key questions
What to take away
What is Mollick’s main advice?
Use AI to understand it and decide where it is useful in work and daily life. Keep skills that help you evaluate the technology.
What does an AI agent do?
It can be assigned a task and carry out multiple steps, rather than responding only to each prompt. Reliability varies by task.
Does this show AI will eliminate jobs?
No. The cited 71% is a measure of public expectation, not evidence of future job losses.
Can using AI prevent its harms?
No. Individual learning may help people adapt, but it does not resolve employment, safety, security, or fairness concerns.
Preparing for AI at Work
The argument matters because AI is moving from a tool people consult to systems that may carry out multi-step tasks. That shift could affect how work is organized and which tasks people perform, while also making administrative chores easier. The consequences will depend on how companies deploy the systems, what they can do reliably and whether workers retain meaningful control.
Public anxiety is part of that picture. A Pew Research Center study published in August, cited in the source article, found that 71 percent of Americans thought AI would lead to fewer jobs over the next two decades. That was seven percentage points higher than the result two years earlier. The figure measures respondents’ expectations, not a forecast or a count of jobs already lost.
Mollick’s recommendation to experiment is therefore best understood as one way for individuals to build familiarity, not as a substitute for decisions by employers and policymakers. Knowing how a tool works may help someone evaluate its usefulness, but it cannot by itself prevent displacement, correct flawed systems or determine who receives the gains from automation.
Products Worth Considering
Bose QuietComfort Wireless Noise-Cancelling Over-Ear Headphones, Black
DJI Osmo Pocket 3 Creator Combo Vlogging Camera with Mic 2, 1 CMOS, 4K/120fps, 3-Axis Stabilization, 128GB Card
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
As an affiliate, we earn on qualifying purchases.
From Chatbots to Agents
The source describes a period of sharply contrasting developments around AI. It cites examples of agents handling reservations and shopping, and an Apple announcement about an AI-ready home display. It also points to concerns about employment, a reported increase in AI spending and fears that systems could be misused or create security problems. These examples are presented as a snapshot of the debate; the supplied material does not independently verify each claim or give enough detail to assess them.
Mollick’s work focuses on how people use generative AI. He is the author of Co-Existence: The Next Phase of AI and writes the newsletter One Useful Thing. His central premise, as presented in the interview, is that people need practical experience with AI to understand the technology’s changing capabilities. The discussion also acknowledges that adapting to AI does not remove the underlying social and economic questions.
“You need to use AI in order to understand it.”
— Ethan Mollick, associate professor at the Wharton School
Limits of AI’s New Role
The interview does not establish how often AI agents complete tasks successfully, how much supervision they require across different settings, or whether the productivity gains reported by some coders are typical. The subscription-cancellation example is Mollick’s account, not a controlled assessment of an agent’s reliability.
It also remains unclear how quickly agents will become common in workplaces and households, which jobs or tasks will be affected, and how any gains will be distributed. The cited survey records what Americans expect about jobs over the next two decades; it does not show what will happen. Mollick’s advice to learn by using AI does not settle these questions or address all possible harms.
How People Can Test AI
Mollick’s immediate recommendation is for people to try AI on tasks where they can check the result, learn how the system behaves and retain skills they may still need. The interview does not announce a formal program, a deadline or a policy proposal. It presents this as a way for individuals to prepare while AI tools change.
For workers and employers, the next practical test is whether agents can perform useful tasks accurately, securely and with appropriate human oversight. More evidence about reliability, workplace adoption and job effects will be needed before broad claims about AI’s benefits or costs can be treated as settled.
Key Questions
What is Ethan Mollick’s main advice about AI?
Mollick argues that people should use AI to understand it and decide where it is useful in their work and daily lives. He also warns against losing skills that help people understand or evaluate technology.
What does an AI agent do?
An AI agent can be assigned a task and carry out multiple steps, rather than responding only to each prompt in a back-and-forth chat. Its reliability and need for supervision can vary; the source does not establish how well agents perform across tasks.
Does the article show that AI will eliminate jobs?
No. It cites a Pew survey in which 71 percent of Americans expected AI to lead to fewer jobs over the next two decades. That is a measure of public expectation, not proof of future job losses.
Can using AI prevent its harms?
No. Mollick presents familiarity with AI as a way for individuals to adapt and make informed choices. It does not resolve concerns about employment, safety, security or how the benefits of AI are shared.
Source: Vox
