TThe Diary of a CEO
← All episodes
12 May 2025

AI AGENTS EMERGENCY DEBATE: These Jobs Won't Exist In 24 Months! Containment Has Failed, We Must Prepare For What's Coming!

6Frameworks
14Insights

Frameworks in this episode

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Hot Take· 4

Hot Take42:00

The Jobs Masad Expects AI to Hit First

Amjad Masad predicts that highly routine computer work is most exposed over the next few years, citing quality assurance, data entry, and text-or-image production. He also cautions that adoption in healthcare is harder to predict because regulation and public willingness can slow deployment, so the timing remains his forecast rather than an established outcome.

  • Routine click-and-type workflows are presented as the most exposed
  • Text-in, artifact-out work is described as being at risk
  • Regulation and adoption can delay technical substitution
  • The timeline is Masad's prediction, not a confirmed labor-market fact

if your job is as routine as it comes, your job is gone in the next couple years

Amjad Masad · 42:30
#jobs#automation#regulation
Hot Take55:30

Careers May Shrink From Decades to Months

Priestley predicts a “high velocity economy” in which careers may form around short projects rather than a single 45-year arc. In his account, small teams build and launch something, are disrupted, and then regroup around another opportunity, making repeated transitions a normal part of working life.

  • Priestley contrasts long careers with projects lasting months or a few years
  • Teams may repeatedly form, launch, disband, and recombine
  • The claim is a forecast about how work could change

we're going to go into a high velocity economy

Daniel Priestley · 55:30
#careers#future of work#teams
Hot Take1:59:30

The Panel's Strongest Positive Case for AI

Weinstein and Priestley identify education and healthcare as major areas where AI could extend scarce expertise. They suggest it might help competent educators reach more learners and improve access to medical guidance or analysis where professionals are unavailable. These are prospective benefits discussed by the guests, not validated medical claims or advice to replace clinicians.

  • AI could help scarce expert teachers reach more pupils
  • Priestley expects potential gains in access to healthcare expertise
  • Weinstein also imagines adaptive governance based on outcomes
  • The discussion is aspirational and does not establish clinical safety or effectiveness

The big ones are healthcare and education.

Daniel Priestley · 2:01:30
#healthcare#education#ai benefits
Hot Take2:17:30

The AI Pleasure Trap That Could End Exploration

Masad imagines individualized virtual environments that simulate relationships, work, and struggle so convincingly that people lose reasons to leave. He connects this speculative outcome to one proposed answer to the Fermi paradox: advanced civilizations may turn inward toward engineered pleasure rather than outward toward exploration. Weinstein adds the simpler possibility that powerful technology enables self-destruction.

  • AI and VR could create private worlds tailored to each person
  • Simulated struggle might satisfy needs that currently drive real-world action
  • Masad presents inward retreat as one speculative answer to the Fermi paradox
  • The scenario is explicitly hypothetical

everyone is living in their own created universe

Amjad Masad · 2:18:00
#virtual reality#fermi paradox#pleasure

Explainer· 5

Explainer19:30

What Happens When Cognitive Labor Becomes Almost Free?

Daniel Priestley compares current AI capability to discovering a vast remote workforce with advanced qualifications, broad language skills, constant availability, and very low cost. He argues that even if AI remained near its present level, that supply shock could force society to rethink work, income, and meaning.

  • The analogy treats AI as an enormous new supply of remote cognitive labor
  • Very low labor cost could disrupt many internet-delivered jobs
  • Priestley says the social implications extend beyond productivity

it's almost as if we've just invented a new continent of remote workers

Daniel Priestley · 19:30
#labor#automation#economics
Explainer1:17:00

Hypernovelty: When Change Outruns Human Adaptation

Weinstein defines hypernovelty as change occurring faster than humans can adapt. He argues that people increasingly grow up in one environment and reach adulthood in another, creating mismatches across multiple areas of life. His test for AI is whether it eventually stabilizes human conditions or accelerates repeated disruption.

  • Hypernovelty is defined as change outpacing adaptation
  • Developmental lessons may no longer fit adult conditions
  • Weinstein links persistent mismatch to harm across several domains
  • He asks whether AI produces one transition or endless event horizons

Hyper novelty is the fact of the rate of change outpacing our capacity to adapt to change.

Bret Weinstein · 1:17:30
#hypernovelty#adaptation#social change
Explainer1:21:30

AI Tutoring Meets the Case for Learning Through Reality

Masad says research on education identifies one-to-one tutoring as an unusually strong intervention and argues AI could make personalized tutoring widely available. He and Bartlett describe adaptive explanations for children, while Weinstein counters that direct interaction with physical systems offers feedback no teacher needs to arbitrate. The research magnitude is reported as Masad's claim and is not independently established by the transcript.

  • Masad presents one-to-one tutoring as a high-impact educational intervention
  • AI can vary pace, questions, and analogies for an individual learner
  • Weinstein prefers concrete tasks where reality supplies success or failure
  • The panel considers combining AI guidance with robotics and physical projects

that is one-on-one tutoring

Amjad Masad · 1:22:00

I like to let physical systems tell the person when they've succeeded or failed.

Bret Weinstein · 1:24:30
#education#tutoring#learning
Explainer1:50:00

Weinstein's AI Risk Map Goes Beyond a Malevolent Machine

Weinstein ranks deliberately malevolent AI as the risk he worries about least, then describes misaligned agents, damage to shared reality, economic and social disruption, and escalation toward war. His emphasis is that harmful outcomes do not require a conscious machine that wants to attack humanity; human incentives, misunderstanding, and synthetic media may be sufficient.

  • Malevolent AI is presented as the least concerning item on his list
  • An agent can cause harm by misunderstanding a goal
  • Undetectable deepfakes can undermine confidence in evidence
  • Economic abuse may empower actors who ignore social constraints
  • AI-driven instability could interact with pre-existing weapons

an AI can be misaligned not because it has divergent interest but because it just misunderstands what you've asked it

Bret Weinstein · 1:50:30
#ai risk#deepfakes#misalignment#war
Explainer2:10:00

Why AI Containment Can Create Its Own Authoritarian Risk

Masad identifies autonomous weapons as a central fear, describing facial recognition, drone swarms, geopolitical competition, and domestic repression as possible uses. He also argues that containing broadly available AI may require intrusive surveillance of computing and ideas, creating a second danger: controls intended to stop harmful systems could themselves become oppressive.

  • Autonomous drones could lower the cost of targeted violence
  • Geopolitical competition may reward speed and corner-cutting
  • The same tools could be used by states, gangs, or commercial actors
  • Masad argues that total containment could require draconian surveillance
  • The segment presents risks and scenarios, not claims that each outcome is occurring

what really scares me is autonomous weapons

Amjad Masad · 2:11:00

You need total surveillance on compute on people's computers on people's ideas

Amjad Masad · 2:17:00
#autonomous weapons#containment#surveillance

Story· 1

Story1:51:30

Bartlett Says Deepfake Scams Are Already Exploiting His Audience

Bartlett says his team repeatedly reports deepfake ads that imitate him, direct viewers to Telegram groups, and seek payment or card details. He describes victims contacting him after losing money and says the fraud also appropriates his credibility. The segment is a first-person account from the host, not an independently audited estimate of prevalence.

  • Bartlett reports repeated fake ads across X and Meta
  • The scams allegedly use his face and voice to direct followers elsewhere
  • Removing one batch does not stop new versions appearing
  • The damage includes both victim losses and stolen reputation

It's a video of me basically asking you to come to a Telegram group where people are being scammed

Steven Bartlett · 1:52:30
#deepfakes#scams#identity

Q&A· 1

Q&A1:07:00

Income Support Does Not Automatically Replace Meaning

The panel separates material support from the psychological role of useful work. Masad argues that people want to be useful and would not necessarily be satisfied by universal basic income, while Weinstein says meeting physiological needs alone would not provide something meaningful to strive for. These are philosophical and psychological views offered by the speakers, not settled clinical findings.

  • Job loss can remove status, identity, and a sense of usefulness
  • Basic income could address needs without resolving purpose
  • The panel calls for deliberate thought about meaningful post-work lives
  • The discussion presents speaker judgments rather than medical conclusions

we want to be useful

Amjad Masad · 1:07:00

We have to have things to strive for

Bret Weinstein · 1:09:00
#meaning#ubi#job displacement

Takeaway· 3

Takeaway57:30

When Building Gets Cheap, Distribution and Humanity Gain Value

The panel argues that cheap software creation weakens capital and engineering headcount as defensive moats. Steven Bartlett suggests an existing audience becomes more valuable when many people can build, while Priestley and Weinstein expect authenticity and proof of humanity to gain value as synthetic content becomes easier to produce.

  • Ideas and development resources may become weaker moats
  • An owned audience can provide built-in distribution
  • The panel expects authentic human performance to become more valuable
  • Anything easily faked may be devalued

the moat is distribution

Steven Bartlett · 58:00

authenticity is going to become the coin of the realm

Bret Weinstein · 58:00
#distribution#moats#authenticity
Takeaway1:36:30

Education May Need to Last an Entire Working Life

Priestley argues that education built around a fixed set of job skills is poorly matched to repeated technological disruption. He proposes treating people as lifelong learners who move between work and re-education, with more emphasis on tools, coaching, judgment, and values than on memorizing facts for one career.

  • Narrow skills may age faster in a volatile labor market
  • Workers may repeatedly move between opportunities and education
  • Priestley favors tools, coaching, mentoring, judgment, and values
  • The proposal expands education beyond childhood and early adulthood

has to be a lifelong learning approach

Daniel Priestley · 1:36:30
#lifelong learning#education#reskilling
Takeaway2:03:30

Why Priestley Thinks Five-Person Teams Can Tackle Big Problems

Priestley argues that a small group united around a meaningful problem can use AI tools to achieve far more than a comparable team could previously. He predicts short, intense cycles in which teams build, scale, and then move together or separately to another problem. The scale and earnings he describes are optimistic projections rather than guaranteed outcomes.

  • Shared commitment to a meaningful problem comes before tool leverage
  • Priestley focuses on teams of roughly five to ten people
  • AI may let small teams build and scale solutions faster
  • Teams can regroup after a project rather than becoming permanent organizations

small teams have infinite leverage now

Daniel Priestley · 2:03:30
#small teams#entrepreneurship#leverage