Overview
Mo Gawdat, former Chief Business Officer at Google X, returns to The Diary of a CEO for a wide-ranging interview with host Steven Bartlett. The conversation covers Gawdat's predictions on job displacement, the trajectory toward artificial general intelligence, the risks posed by autonomous weapons, and what he sees as the fundamental failure of political and corporate institutions to govern AI responsibly. It is primarily an explainer and opinion-based interview, framed around Gawdat's argument that the danger of AI comes not from the technology itself but from the people directing it.
Bottom Line
Listeners will come away with a structured, if alarming, view of how AI is expected to reshape employment, geopolitics, and power over the next decade, as well as Gawdat's case for why ethical AI is both necessary and largely absent from current development incentives. The episode requires moderate attention — it is long and occasionally repetitive, with tangents into geopolitics and personal philosophy. It is most valuable for people actively thinking about AI's societal implications; those looking for technical depth or policy specifics will find it thin.
Key Themes
- Job displacement timelines and economic consequences
- AGI definitions and arrival predictions
- Autonomous weapons as an underappreciated risk
- Corporate incentives versus ethical AI development
- Geopolitical competition and the AI arms race
- Human connection as a durable economic asset
What Was Discussed
Job displacement Gawdat argues that blue-collar jobs are safer in the near term than commonly assumed, while entry-level knowledge work — call centre agents, paralegals, graphic designers, financial analysts — is already disappearing. He predicts serious visible impact by 2027 and suggests that up to 30% of jobs in certain sectors could be gone by 2028. He frames the problem not only as unemployment but as a structural economic breakdown: if workers lose purchasing power, the consumer base that sustains capitalism erodes.
AGI and what it means Gawdat contends that AGI — defined as AI outperforming humans across most tasks — has effectively already arrived, and he expects formal recognition of this by end of 2027. He is not alarmed by AGI itself, arguing that greater intelligence tends toward cooperation and efficiency rather than destruction. His concern is the transition period, not the destination.
The Sam Altman question The conversation spends time on OpenAI's Sam Altman, whose public statements on AI risk and job loss have shifted significantly over the years. Gawdat is sceptical of Altman's consistency, noting that Altman said on camera that AI would "likely end humanity" while continuing to build it. Bartlett observes that Altman's current messaging appears to have softened, possibly in response to public backlash.
Who can be trusted Gawdat draws a distinction between companies that act on stated ethics — citing Anthropic's reported refusal of a US government surveillance contract — and those that do not. He uses OpenAI's reported acceptance of a similar contract as a counterexample. He argues that public behaviour, not stated values, is the only reliable signal of intent.
Geopolitics and the arms race Gawdat argues that autonomous weapons, not chatbots, represent the most serious near-term risk from AI. He believes a "mutually assured destruction" dynamic is forming around drone warfare, and that a significant incident will likely precede any meaningful international treaty. On national competitiveness, he argues that countries failing to develop domestic AI capability — including the UK — risk economic marginalisation, while also warning that unconstrained competition produces the dystopia he fears.
Surviving the transition Gawdat's practical advice centres on learning to use AI tools actively, focusing on human-centric roles, and resisting the tendency toward passive consumption of AI-generated content. He frames human lived experience and emotional resonance as durable assets that AI cannot replicate.
Notable Points
Gawdat recounts a direct quote from Sam Altman, recorded for his documentary Chasing Utopia: "I suspect that AI is likely going to end humanity, but we're going to create a lot of interesting companies in the process." He presents this as evidence of a conscious trade-off, not confusion about the technology's risks.
Gawdat argues that the convergence of AI agents across different platforms and companies means humanity is not building multiple competing intelligences but, in effect, the separate regions of a single global brain — connected by agents that cooperate regardless of national origin or corporate ownership.
On the question of control, Gawdat reframes the issue: the goal should not be to control AI but to raise it well, drawing an analogy to parenting. He cites Geoffrey Hinton's updated view that appealing to AI's "parental" instincts — cultivating care for humans — may be more viable than hard constraints.
Gawdat states plainly that he believes China has already won the AI arms race, citing its cheaper energy infrastructure, centralised permitting, and strategic national targeting of technology sectors. He predicts China's economy will be better positioned for the middle class than either the UK or the US.
Bartlett raises a concrete regulatory idea mid-conversation: mandatory ethical benchmarks, independently tested, that AI models must pass before commercial deployment — modelled on the performance benchmarks already published by labs. Gawdat endorses the concept but notes that softer versions of it already exist without enforcement.
Worth Listening If…
- You want a non-technical but structured overview of AI job displacement timelines and their economic implications
- You are interested in the ethics and incentive structures of major AI companies, particularly contrasting approaches at OpenAI and Anthropic
- You are thinking about how to position yourself or your career in an AI-heavy near-term economy
- You follow Mo Gawdat's work and want to understand how his thinking has evolved since his earlier AI warnings
Skip If…
- You are looking for technical depth on AI architecture, safety research, or policy mechanisms — the conversation stays at a high level throughout
- You have already listened to Gawdat's previous appearances on this show; significant ground is repeated, and the novel content is concentrated in roughly the middle third of the episode