If AGI Is Coming, Why Invest in Anything?

Ethan Mollick, who’s become something like AI’s David Attenborough, recently posed a question to investors:

If you believe AGI is close, how do you justify backing a startup with a 5–10 year horizon?

It’s a fair challenge, and one that deserves a better answer than: “We believe in founders.” I make no claims to be a macroeconomic expert, but as someone who’s invested through the internet and mobile revolutions, and who’s now luckily appeared on the last four Midas Lists, I feel somewhat qualified to opine about the future of AI and VC. Here’s why I remain bullish:

AI Takes Time to Metabolize

As professor Mollick has previously noted, even if research on models stopped today, it would probably take a decade or more for the full impacts to be felt. It took a decade for smartphone penetration to reach 70% in the US. AI’s business model layer is still in beta and tooling, interfaces, workflows, and norms are far from settled. Venture capital doesn’t fund inventions, it funds adoption, a process that is full of second-order effects that create seemingly endless opportunities for entrepreneurs.

Stubborn Attachments to IRL Experiences

Humans are also just stubbornly attached to IRL experiences. People want to partake in live sporting events as much as sports betting. AI can out diagnose a medical resident, but someone still needs to change bed pans and excise tumors. DaVinci robots are making progress in this realm, but I don’t see a world in the next 10 years where people outsource dating, dining, childbirth, or décor entirely to machines.

Legal Challenges

We were proud investors in Cruise Automation in 2014 and in 2016 it was acquired by GM. Nearly a decade later, Level 5 autonomy remains out of reach. Even Waymo, the best-in-class offering, only operates in a few cities.

We’re living in a moment where regulators aren’t entirely sure how to handle AI. Many would like to ban it, but the potential geopolitical risks of falling behind in the AI race leaves many unwilling to risk that draconian decision.

The sheer novelty of some of the legal questions surrounding AI make it a challenge to even analogize properly. This ambiguity creates opportunities for founders who can take bold risks while incumbents fret about regulators.

Energy isn’t Infinite

We may soon hit a wall where hard Energy limits crimp the era of AI abundance. I have no doubt entrepreneurs will find ways to solve the energy problem over time, and I’ve made a recent investment in this space, but energy production is a slow process and may give us a few more decades of relative normalcy

Venture Thrives Not on Technology, but Human Weirdness

Maybe this is hopelessly old-fashioned, but I believe human beings are more than language models. We are embodied intelligences whose decision-making relies not just on logic, but from gut bacteria, dreams, hormones, grief, and hope. If LLMs can unlock more productivity, and if they don’t end up starting a nuclear war – some big ifs – you have to have faith that human beings will find new ways to create value. It is very likely that an AI will soon make better decisions about oncology and produce overtures that would make composers weep, but many believe human beings carry a divine spark that no AI can ever fully capture.

The things we do may be weirder or more wonderful than we currently expect. Imagine telling a worker at the Triangle Shirtwaist Factory that one of their descendents would make money playing video games online.

This is not to say there won’t be massive disruption and negative side effects and they should be weighed by policy makers. Software engineers may soon experience some of the dislocations experienced by steelworkers in their parent’s generation. These are problems we have to solve collectively, but I remain confident that venture as an asset class will play a part in funding whatever these solutions turn out to be.

The VC playbook, as flawed as it is, started out as a method to fund whaling ventures. Yes. It has helped enable mass manufacturing, automobiles, air travel, telecommunications and the full scale social tumult that has followed in their wake. AI is a massive wave, but it isn’t the white whale.