Preparing for the worst
With the recent drama over at OpenAI, a fire has been lit under my rear.
The Drama: Sam Altman was relieved as CEO of OpenAI because the board wants to focus more on building safe AGI for humanity, and they felt Sam was not being forthcoming. Other key employees resigned in solidarity. The board seemed to consider a reversal where they would resign and Sam might come back, but now it looks like they are staying while Sam and others are moving over to OpenAI’s biggest investor, Microsoft. And now it looks like the staff of OpenAI are threatening to follow Sam and friends to Microsoft if the board doesn’t resign.
“If you strike me down, I shall become more powerful than you can possibly imagine.”
Obi-Wan Kenobi, Star Wars: A New Hope
Ouch. I built my newsletter concept around the idea that ChatGPT would continue to work as well if not better than it does now. I never considered the possibility it could all go away.
Not that this means ChatGPT is going anywhere. It’s just a reminder that nothing is certain. So I’m focusing again on developing as many exercises as I possibly can so I’ll have time to transition if OpenAI falls apart.
Reminds me of the people who staked their claims on Clubhouse when that was the shiny new thing.
Thankfully I have other options if Chatty dies, but it’s still my favorite.
I think Clubhouse was an interesting example of a cursed problem. When it was small, the elites started using it, making it seem like an exclusive club. The more popular it became, the less valuable the experience was.
@Winkletter I’m watching all the memes and tweets play out on Twitter. It’s like an entire season of HBO’s Silicon Valley, only better and in real life haha. Such a wild week. I wonder if chatgpt will still be running next month, or will everyone move to Azure OpenAI…
Which open source models are you experimenting with? I’m playing with codellama-34b. Pretty decent and matches chatgpt 3.5
HBO’s Silicon Valley still holds the record for the greatest dick joke of all time, which I believe occurred in the season 1 finale.
@jasonleow OpenAI screwed up by letting their board get too small.
There are so many open source models now, I need an AI to help me figure out what to try next. When I was more hands on earlier this year I wasn’t that impressed. But I’m hoping they are much better now.
@therealbrandonwilson that d*ck joke was EPIC. *pun intended
@jasonleow Thanks for the tip. I’m going to try it out now. I like that it can tell me which models should work on my system.