Large companies have different levels of risk tolerance
Yet another fragment of the internet talking about large-language models
After working at a large tech company for the past five years, I think I’ve truly internalize the concept that large companies are risk averse as a means of survival. For most multi-billion dollar conglomerates, the only things that can make that company stop existing are:
Getting regulated out of business
Getting sued into oblivion (possibly due to government regulations)
Playing a large role in a finance-related scandal
Big companies tend to be structured in ways to minimize the chances of these large-scale events happening. If government regulation was the most plausible existential threat to my existence, I’d spend tens of millions of dollars per year on lobbyists too.
Startups on the other hand have very different existential threats. Notably, most of them are going to run out of money in a length of time measured in days or months. Sure, you could end up like Aereo and have a US Supreme Court ruling put you out of business. But most startups fail in a more traditional way: burn through too much money and fail to find more of it. For startups, inaction is the leading cause of death.
The interesting bit that spurred me to write this post is that different large tech companies have materially different appetites for risk. Consider Microsoft with a market cap of $1.9T and Alphabet with a market cap of $1.2T.
Alphabet made a big announcement about Bard, someone pointed out that it made a factual mistake in the initial advertisement, the stock price dipped 8%, and Alphabet has been silent about Bard ever since.
Microsoft on the other hand released ChatGPT + Bing, it’s also making factual errors, and it’s even started comparing fairly reasonable people to Hitler.
Either one of these companies could open itself up to massive lawsuits if their AI says something it shouldn’t. It could lead to widespread user distrust or draconian regulations by lawmakers who understand the technology even less than AI researchers. And despite these risks, Microsoft decided to keep their AI experiment out in the world while Google is re-evaluating their timing and messaging.
Why these two companies have different risk appetites for LLMs is an interesting question that I will probably never hear an honest answer for. Perhaps Microsoft views this as a brilliant opportunity to gain search and browser market share. Maybe Google is under increased regulatory scrutiny. Regardless, a company’s size is a factor and not the factor in determining risk appetite.