So who owns large stakes in OpenAI exactly? Seems like a somewhat important question given how important the company is shaping up to be to the future of society.
Clearly Microsoft has a large stake as well. It sounds like equity was distributed to employees, but according to Fortune some was sold: "OpenAI’s other investors include Hoffman’s charitable foundation and Khosla Ventures. Last year, Sequoia Capital, Tiger Global Management, Bedrock Capital, and Andreessen Horowitz reportedly purchased shares from preexisting shareholders in a sale valuing the company at around $20 billion, according to The Information."
https://fortune.com/2023/01/10/microsoft-investment-10-billi...
OpenAI has been selling themselves as a do-gooder kind of project. I think that they're being disingenuous in doing so. They're building just a regular old SV money-spinner.
Every interaction with the public and these AIs I see screams "this was not ready for general consumption"
There are certainly huge positives, but do you really feel something like Facebook is a net positive? Facebook, which intentionally stoke(d/s) genocide? Genocides have existed before Facebook, yes, but so did communication and racist relatives.
[1] https://www.amnesty.org/en/latest/news/2022/09/myanmar-faceb...
Code for America [https://codeforamerica.org/]
SFTech4Good [https://www.sftech4good.com] - a newsletter and monthly meetup in SF for orgs/people interested in technology and social impact
There are also social-cause startups (which obviously have to be monetized by some funding source, typically govt, grant or ESG).
Also, Yahoo (RIP) was by reputation as community-minded and pro-open-source as a for-profit business citizen can get.
As someone who went through Yahoo acquisition and total bungling of tumblr, where community feedback fell on deaf ears, that's news to me. (Granted, the next owners were the ones who immediately managed to drop the userbase by like 80%, so maybe you're not totally wrong)
(IIRC Yahoo culture had the reputation for being very friendly to employees open-sourcing their code, and not aggressively pursuing BS patent suits; this was different to most of MAANG + telecomms.)
Avoid:
- Thiel and Tiger (Coleman III) - Political meddlers
- Khosla - beach-hoarder and waster of taxpayer dollars
- Social Discovery Group - Russia
https://twitter.com/elonmusk/status/1639142924985278466?s=46...
Which makes Microsoft the majority shareholder. Which means that for all intents and purposes, we can consider OpenAI as a Microsoft venture now.
It's definitely a terrifyingly powerful company, but "future of society" seems a little strong.
I think it's the future for spam, writing rote emails/documents/code, disinformation, plagiarism, lying, cheating, and impersonation, but those are subsets of society -- not all of society.
The answer was useless btw. Just a coarse, high-level, unactionable summary of what we had already talked about. Dressed up in a few nicely-worded paragraphs.
I wonder if this is gonna be a new pain in my ass. Wait until managers start using it to disagree with my estimates - I might blow my top lol.
1. Answer student questions and effectively act as a tutor. Duolingo is already using it like this, and ChatGPT can of course do it directly as well.
2. Make natural language interfaces to APIs easily. Look at the ChatGPT plugins announced yesterday for examples.
3. Provide basic customer support. Ideally, it could answer most common and basic questions and possibly even fix common problems via a plugin. Then the actual human customer support could step in for more complex problems.
There are endless positive examples like these.
Ask 10 different people "smart" people (define smart however you want), you'll get 10 different answers to these questions. These are all questions an LLM could answer amazingly. Probably a lot better than most humans.
If you don't ask it what 2+2 is or who came to in America in 1875 then you get useful things.
Asking a LLM deterministic questions right now is like asking a calculator what the meaning of life is. If you use the tool for something it's not good at you get unusable answers.
Further, 10 different smart people will give 10 different answers because they have coherent worldviews and biases and proclivities, so by accounting for those, you can work out what the right answer is. Even if ChatGPT was anywhere close to a human expert when it comes to accuracy (what's the error rate in a peer reviewed journal article?) it would still have no coherent worldview or bias to contextualize its statements.
HN seems to want to hand over the keys to the kingdom to basically a string generator. The string generator believes nothing, understands nothing, knows nothing but here we are.
Any intelligence that gpt4 shows is an emergent property. Humans are the ones reading GPT’s output and imputing meaning to it.
Reminds me of astrology and mass hysteria - people convincing each other to give this new oracle a chance because they personally have seen value in its ramblings.
What is your error rate? What is my error rate? All of this stuff is unknown because we don't have counter factuals and we don't think of the world in this way (Did you order the correct food at dinner? Did you wash your clothes at the optimal time?)
To me you're thinking of it as a classical deterministic (binary) computer rather than a probabilistic thing. It's not an oracle, or a miracle, or anything other than some thing that gives useful information some percentage of the time. If something has to be right 100% of the time for it to be useful, or even 60% of the time, then the world is missing out on a lot of value.
Investors are right ~51% of the time, startup founders in the aggregate are right ~10% of the time, a great batting average is ~30%, etc.
One big area where it does poorly right now is math. But they just announced a ChatGPT plugin for Wolfram, which I expect will make it very good at math. Wolfram also has a large database of curated information to draw on.
Technology improves over time. GPT is still new and improving quickly. What it does now isn’t perfect, but it is still incredible.
My guess is stuff like math, where you can fairly easilly verify the factuality of ChatGPT's answers, is an area where you could certainly see progress. More general stuff like history, where it's important to have a really firm grasp of facts, inutition, and nuance, ChatGPT will likely be hard to improve, and worse, much harder to verify. Worse, these things can be insiduous: if you've learned something straightforwardly wrong, it corrupts future conclusions drawn from that erroneous premise.
Need to do math? Use the Wolfram plugin.
Need to have hard facts from reliable and citable sources? Use a plugin that queries databases like Arxiv. The LLM could give you links to sources and provide quotes from those sources to support its reasoning.
/-s