Sam Altman goes before US Congress to propose licenses for building AI

My gut feeling is that the majority of AI safety discussions are driven by companies that fear losing their competitive edge to small businesses. Until now, it's been challenging to grow a company beyond a certain size without employing an army of lawyers, human resources professionals, IT specialists, etc. What if two lawyers and an LLM could perform the same work as a legal department at a Fortune 500 company? The writing is on the wall for many white-collar jobs, and if these LLMs aren't properly regulated, it may be the large companies that end up drawing the short straw.

How many of Microsoft's 221k employees exist solely to support the weight of a company with 221k people? A smaller IT department doesn't need a large HR department. And a small HR department doesn't file many tickets with IT. LLM driven multinationals will need orders of magnitude fewer employees, and that puts our current multinationals in a very awkward position.

Personally, I will be storing a local copy of LLaMA 65B for the foreseeable future. Instruct fine-tuning will keep getting cheaper; given the stakes, the large models might not always be easy to find.

I keep seeing AI leaders looking outward and asking for 'someone else' to regulate their efforts, while they're accelerating the pace of their efforts. What's the charitable interpretation here? Elon Musk, too, has been warning of AI doom while hurriedly ramping up AI efforts at Tesla. And now he keeps going about AI doom while purchasing thousands of GPUs at Twitter to compete in the LLM space. It's like "I'm building the deathstar, pls someone stop me. I won't stop myself, duh, because other ppl are building the deathstar and obviously I must get there first!"

Remember that popular recent post about OpenAI not having a moat? Well it looks like they are digging one, with a little help from the government.

We need to MAKE SURE that AI as a technology ISN'T controlled by a small number of powerful corporations with connections to governments.

To expound, this just seems like a power grab to me, to "lock in" the lead and keep AI controlled by a small number of corporations that can afford to license and operate the technologies. Obviously, this will create a critical nexus of control for a small number of well connected and well heeled investors and is to be avoided at all costs.

It's also deeply troubling that regulatory capture is such an issue these days as well, so putting a government entity in front of the use and existence of this technology is a double whammy — it's not simply about innovation.

The current generation of AIs are "scary" to the uninitiated because they are uncanny valley material, but beyond impersonation they don't show the novel intelligence of a GPI... yet. It seems like OpenAI/Microsoft is doing a LOT of theater to try to build a regulatory lock in on their short term technology advantage. It's a smart strategy, and I think Congress will fall for it.

But goodness gracious we need to be going in the EXACT OPPOSITE direction — open source "core inspectable" AIs that millions of people can examine and tear apart, including and ESPECIALLY the training data and processes that create them.

And if you think this isn't an issue, I wrote this post an hour or two before I managed to take it live because Comcast went out at my house, and we have no viable alternative competitors in my area. We're about to do the same thing with AI, but instead of Internet access it's future digital brains that can control all aspects of a society.

If AI is to be a consumer good—which it already is—it needs to be regulated, at the very least to ensure equal quality to a diverse set of customers and other users. Unregulated there is high risk of people being affected by e.g. employers and landlords using AI to discriminate. Or you being sold an AI solution which isn’t as advertised.

If AI will be used by public institutions, especially law enforcement, we need it regulated in the same manner. A bad AI trained on biased data has the potential to be extremely dangerous in the hands of a cop who is already predisposed for racist behavior.