robbomacrae

For the brand, the IP, and the fact it would be pennies on the dollar.

How would such a merger be good for AI safety? A combined OpenAI-Anthropic would collectively own 6 out of the top 6 LLM's on the chatbot arena leaderboard [0] which would seem to be somewhat monopolistic. This would result in less choice for us developers.

Isn't it better to have the top two AI firms both upholding AI safety as a core principle?

To me this weakens the original argument they had for sacking Sam in the first place.

[0] https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboar...

Honestly I think there was an offer and Sam verbally accepted.. maybe even signed. That doesn't mean it will happen and if all this is the case.. then it should be fine by SEC because there wasn't technically a lie.

The most damning part of all for the remaining board is that a week ago the thought of OpenAI dying would have been unthinkable but now people are genuinely worried it might happen and what the implications are. Even if safety was the genuine issue.. they should have planned this out a lot more carefully. I don't buy the whole "OpenAI destroying itself is in line with its charter" nonsense.. you can't help guide AI safety if you don't exist.

"We are really excited to work with Sam, OpenAI 2.0, the team and the new board of directors who we strongly believe will achieve great things in partnership with Microsoft. We've already signed new deals to strengthen this relationship and more details will be coming soon."

- Satya (probably)

This might be my favorite comment I've read on HN. Spot on.

Being able to watch the miss steps and the maneuvers of the people involved in real time is remarkable and there are valuable lessons to be learned. People have been saying this episode will go straight into case studies but what really solidifies that prediction is the openness of all the discussions: the letters, the statements, and above all the tweets - or are we supposed to call them x's now?

I doubt Sam was personally that involved in the mechanics of the models.

But there is an alternative scenario.. in order for Microsoft to avoid loosing any momentum they might offer Altman an insane amount of money to become the John Giannandrea of Microsoft and bring as many of his recent colleagues with him. And for Altman this might be the easiest way to not loose ground as well with Microsofts patents and license agreements.

Yes that's my take. I think its good enough for most use cases but if you really wanted that random effect you might consider replacing the lines using something like an ipad and pencil or stylus.

Shameless plug but this is exactly how I put together the graphs in my blog post[0]. It's very easy to use and surprisingly versatile although I did add a bit of photoshopping to get things exactly as I wanted them.

[0] https://summer.ai/blog ChatGPT should be your first cofounder

Well that hardly seems ethical...

Want to bet? Altman has been building himself as the public face of AI and won't let that go so easily. It's not like his world coin is taking off...

And non competes can be as airtight as you like.. they are completely unenforceable in California which is where OpenAI's HQ is based and where Sam Altman lives.

First the loyal will jump ship. Followed quickly by the mercenaries who see the chance to join a new rocket ship from the ground up. Then as OpenAI's shares tank on the secondary market the rest will decide they've seen enough of their paper money burn and cash out.

OpenAI will survive but its going to be a much smaller company and a much smaller valuation.

As for Microsoft I'm guessing one of the strings Nadella was pulling was threatening to revoke credits, resources and even use by his teams and I'm sure he would be interested in investing in whatever Altman starts next and dedicating those now spare machines to the new enterprise.

I think you're mistakingly thinking Mira Murati was on the board. That is not the case. The board is Adam D'Angelo, Helen Toner, Ilya Sutskever, and Tasha McCauley after Sam and Greg were removed. Mira Murati was not involved except for the fact she was made the initial replacement. For all we know she had no information on what happened when she was asked to step in.

There is an enormous YC bias here. ShowHN posts from new YC startups are upvoted instantly whilst most from outside are ignored - the ones you see are the exception. The HN rules state you are not meant to play the game of asking people to boost your post but it seems like time and time again they co-ordinate, or allow it, for the insiders.

I'm in this boat. Not for my startup but for side projects I was absolutely pinning my hopes on them unlocking access to tools and relaxing some of their restrictions in the near future.. a future which now seem unlikely.

It does seem kind of crazy how much negativity there is in the replies to this post and while non stand out as particularly misogynistic on their own... the sheer number of doubters does seem damning.

I wonder why you would have doubts about someone you clearly don't know anything about? Before becoming a partner and then CEO of YC, Sam Altman was a founder of Loopt (failed startup). I was surprised at his sudden rise at YC but it seemed to work out. No reason why this shouldn't be the case here.

Did Steve Jobs have any?

Having read the article I do get the feeling that safety is an underlying current in most of her answers. This certainly gives weight to this being a safety first camp vs growth first camp theory. I hope they don't slow down the pace of innovation too much as there are many startups just waiting for a few improvements here or there before something else becomes viable. And given her answer to the following question I am somewhat optimistic:

Steven: "[...] By putting things like GPT-4 in the wild, it’s almost like you’re forcing the public to deal with those issues [...]"

Mira: "[...] especially as the technological progress continues to be so rapid. It's futile to resist it. I think it's important to embrace it and figure out how it's going to go well."

Unlikely.. if I understand the situation right it's the folks more dedicated to safety that kicked out those more interested in profit. Amodei (Anthropic's CEO) "split from OpenAI after a disagreement over the company’s direction, namely the startup’s increasingly commercial focus." So if anything those who walk now are less likely to be aligned with Anthropic's mission.

I find that bart large 410m [0] parameters does a fine job at summarizing. In Summer AI I alternate between using a copy of that bart large getting hyper-trained on feedback and Chat GPT 3.3 and honestly I don't have a preference between the results.

However, thanks to this article I might revisit the summarization techniques used a try a fine tuned 3.5.

It would be great to see these techniques compared to Chat GPT 4 Turbo.

[0]: https://huggingface.co/facebook/bart-large-cnn

Do you think it is completely out of reach for the AI to follow those rabbit holes automatically and tie in the useful information? Could it not also be personalized to the users knowledge of the subject?

I'm actively working on the first problem. The second is in my todo list.

Hello! I know of your guys work! I try to keep up with all the competitors :) Feel free to reach out to me rob @ summer dot ai and I'd be happy to talk shop.

For anyone else that is interested in this question: I've tried a whole bunch of the TTS services and found that Microsoft and AWS are the best of the standard providers IMHO and these are services that tend to have startup credits available so I use a mix of these two - I try to never rely on just one provider. I've met with the Eleven Labs folks and some of their demo's of the V2 stuff that's coming are really amazing but latency and pricing might rule them out as an option for the time being.

Yes gpt 4. I was fact checking and using places I knew pretty well and would find the occasional error when asking it to describe cities etc and what to do.

Well for itineraries, given that its training data gets cut off at some point, I’d also be worried about it recommending places that are no longer open.

Note that I’m building an app for customers and attempting to build a brand that can be trusted - whereas in your case gpt 4 is probably ideal given you know it’s caveats and can check the important stuff.

I was a bit surprised about their choice of example app. With tour guide/info you need to be re-assured the information you're giving is correct. But asking LLM's to write it for you is a recipe for hallucinations. I built Summer AI to do audio tour guides using old fashioned web scraping and then piping the output through LLM's just to summarize and even then use a couple of extra steps to ensure factual accuracy.

During testing I found any other approach would always inevitably make stuff up.

Not trying to disparage this work. I think it's great that you can build such a thing so quickly. But I wouldn't rely on the info these LLM's provide.

This is pretty much exactly how I've attempted to make Summer AI app work. If I am missing anything from your vision please let me know!

I put together something in a similar vein but attempts to build up an audio tour like experience from the ground up.. old fashioned web scrape then summarized through LLM's: https://summer.ai

Ah I see the creator is in here. Will reach out!

For anyone who thought generative AI wasn't dangerous! I'm sure a lot of the safeguards being developed at OpenAI etc are aimed at stopping this sort of thing.

I tried out some of the captioned messages on ChatGPT and sure enough I received lectures on how "taking another person's life is a severe and irreversible action, prohibited by law and generally considered morally and ethically wrong."

But its a healthy reminder for anyone out there building LLM chatbots to test for this and make sure they don't just blindly encourage a user to follow their chains of thought.

This is a really good argument. One that I haven't heard before. Is there a name for it? It seems to cover some of the backward steps we take when meandering (hopefully) forwards that I for one hadn't considered.

I suspect it's not worth pausing anything over due to the inevitable if we won't they will excuse and also because its part of the re-invention of business that goes with new technology. The things you want that have gone are now an opportunity for someone else to figure out. If I was an entrepreneur keeping my ear to the ground I'd weigh up starting a boutique produce delivery services, a human agency dedicated to dealing with companies on your behalf, a wix plugin market, a book review site etc.

This is potentially a huge deal. Companies are concerned using ChatGPT might violate data privacy policies if someone puts in user data or invalidate trade secrets protections if someone uploads sections of code. I suspect many companies have been waiting for an enterprise version.

I had to go and try this exact line of questioning with ChatGPT because I suspected this might lead to a weakness in it not admitting when it just doesn't have a clue (which would have been my answer)... mind you its a big human weakness/tendency to not admit lack of knowledge.

But the answer was surprisingly candid and yet thoughtful:

""" I can't know for sure why you asked the question about the relationship between "Romeo and Juliet" and "Star Trek," as I don't have access to your personal thoughts or context. However, some potential reasons might include:

Academic Inquiry: You might be exploring themes in literature or media studies and are interested in drawing connections between different works across genres and time periods. Creative Inspiration: If you're a writer, artist, or content creator... """

There were some others but overall I thought the initial disclaimer along with some possible theories approach was spot on and a lot better than my "no clue" knee jerk reaction.