An investor in 50+ AI startups warns against building a “God-like AI” and says companies are “running towards a finish line” without knowing the ramifications
The writer of this essay is an investor and co-author of the annual “State of AI” report
Financial TimesIan Hogarth
Context & Ripple Effects
This April 2023 warning from a State of AI co-author who backs 50+ startups lands one month after fellow VCs began questioning whether generative AI companies attracting huge valuations can actually make money — the same month the hype-cycle comparison to crypto surfaced. It marks the moment the safety argument moved inside the investing class rather than staying with researchers.
Founders across the investor's 50+ company portfolio now hear their own backers argue the race is outrunning its ramifications — a direct challenge to speed-at-all-costs positioning at the labs competing for talent and capital.
The critique gives LPs and public-market allocators a credible internal voice for caution, reinforcing the valuation skepticism already running through generative AI deal-making since March 2023.
Second-order effects
Capital discipline tightens: the caution compounds warnings like Vanguard chief economist Joe Davis's call that the 2024 rush into AI stocks overplays near-term potential and risks a share-price correction, pushing diligence toward revenue durability over capability demos.
Labs racing hardest must spend more effort justifying pace — safety commitments become a fundraising differentiator, splitting the investor base into the advancement-versus-concentration camps documented through 2024.
Third-order effects
If the pattern holds, the industry splits structurally: labs aligned with investor safety pressure converge toward the pause-and-assess posture Anthropic articulated in 2026, while holdouts compete for the capital that remains — concentrating frontier development in fewer hands.
Investor sentiment becomes a de facto governance mechanism ahead of regulation: when even the most plugged-in backers flag unknown ramifications, funding terms and disclosure expectations shift without any regulator acting first.
The trend: AI's most connected investors have evolved from questioning startup valuations to actively shaping how fast frontier labs run, making capital allocation itself a lever on AI safety.
This article on the AI-race is worth a read. I have to say, though, I'm always struck by the almost child-like geopolitical naivety of the solutions that some of these AI experts are coming up with https://www.ft.com/... https://twitter.com/...
Important read on AI from @soundboy, especially as we are not well equipped to grapple with systemic & fat tail risks. God-like AI presents one of the most serious existential threats we have faced as a species. https://www.ft.com/...
Fascinating and important article by @soundboy calling for slowing down of race to AGI in 3 ways: 1. Governments should grill leaders of research labs 2. Employees in those labs should quiz their leaders 3. Time to create a CERN-like entity for AGI https://www.ft.com/...
Ian Hogarth (@soundboy) is an engineer, researcher & long-time investor in AI startups. In this week's @ftmag cover essay, he writes an urgent warning about how dangerously unregulated and poorly understood it really is #longreads https://enterprise-sharing.ft.com/ ...
“I gradually shifted from shock to anger. It felt deeply wrong that consequential decisions potentially affecting every life on Earth could be made by a small group of private companies without democratic oversight” @soundboy https://www.ft.com/...
A very thoughtful piece in FT on governance of AI, by Ian Hogarth https://www.ft.com/... — strengthens the argument I made in my Monday Mint column: The answer is to throw more human brain power at the policy problem.
Saying out loud what many are increasingly but privately fretting over - @soundboy in always classy @FTMag sounding the alarm over the race for an AI that can do anything we can https://www.ft.com/...
One of the best AI essays that I've seen in a mainstream newspaper, from @soundboy Key democratic question - the explicit aim of several AI companies is to create AI better than humans at almost any task. Shouldn't the public get a say on if we want this? https://www.ft.com/... h…
This is a sobering read on our hell-for-leather dash towards more sophisticated AI. It's the classic Jurassic Park conundrum. Just because we can, does that mean we should? (£) https://www.ft.com/...
“My view is that the present and future harms of AI are not mutually exclusive and overlap in important ways. We should tackle both concurrently and urgently” https://www.ft.com/...
Why a prolific investor in AI is now having second thoughts about the dangers of creating a super-intelligence: “God-like AI could be a force beyond our control or understanding, and one that could usher in the obsolescence or destruction of the human race.” https://twitter.com/.…