Another OpenAI founder moves to arch-rival Anthropic Anthropic has hired yet another of OpenAI's founders, this time bringing on Durk Kingma in an unspecified role. This comes after fellow OpenAI co-founder John Schulman left the super lab in August to join Anthropic to "return to hands-on technical work." Kingma revealed his news in a post on X on Tuesday, saying he was …
So, most major founders of an open-source thingy have fled the house to another, non-Redmond-managed one ?
There's a clue there. I'm sure of it.
I mean, it's not like Redmond has ever declared that Open Source is a cancer, now is it ? Oh wait . . .
But is it Open Source at all? "Concerns that Altman and his team weren't being as open as the company name suggests."
It's back to Yes Minister, series 1, episode 1: "getting rid of the difficult bit in the title" and the difficult word there was also "open".
What open source?
Show me the money, err I mean code.
Show me the code for ANY ChatGPT model, bot or service. You can't because none of it is open source.
The 'open' in OpenAI does not stand for open source, it stands for open to taking all the worlds money in an enormous AI scam.
For those who don't think its a scam consider this... Sam AltMan recently admitted that true AI, or whatever he's calling a computer system that's actually intelligent as opposed to his word guessing bots, will take a few thousand days. That's 5 - 10 years. Or the same as predictions for the flying car, cold fusion, quantum computing, etc.
Show me the code for ANY ChatGPT model, bot or service. You can't because none of it is open source. -- Actually the CODE for, e.g., Llama 3 is open source and can be downloaded from Hugging Face. The pre-trained weights, however, are not open source.
But you are right that OpenAI's source code is not open source AI code.
It has nothing to do with open source. Claude isn't open source either. Not in real terms, nor in the sort of open but not fully terms that Facebook used for LLAMA. These people didn't do this for open source. Why they did it is harder to answer. In some ways, though, you can still blame Microsoft for part of it. When OpenAI tried to get rid of Altman, Microsoft was one of the reasons why they failed and he stayed on. Probably not the largest of the reasons, though you're free to think otherwise. If you actually want to know this, try to interpret the vague pronouncements made by all involved in that event. It has more to do with it than open source does.
Not quite bursting yet.
Don't get me wrong, we're definitely in an AI Winter at the moment ... but the momentum of the failed LLM model is still driving suckers investors. Unfortunately, they are slowly beginning to realize that it's all a scam that doesn't work as it says on the tin, and in fact often works exactly the opposite of what it says on the tin. Which might be ::insert villain tunes:: dangerous, in some nebulous, wishy-washy namby-pamby hand waving way that is never fully explained.
So before the bubble bursts, we are seeing another small wave of companies supposedly trying to make LLMs into a kinder, gentler, more woke version ... which is also doomed to failure, of course, because the entire methodology is fundamentally flawed.
Give it another year or two, until the suckers investors run out of cash, and then it'll fade away for another 20 years or so.
Regardless of the poor quality of the LLM's output or how many safety guardrails they manage to bolt on, the fact is the entire AI bubble is based on the wholesale theft of copyright data from the internet. If any form of 'AI' is to emerge after the inevitable crash, the creators of the model must be able to prove permission and licences to use every last byte of training data.
I think this is pretty interesting because there has been a selection of borderline cargo-cult grifters who have been falling forward through the "next big thing" bubbles from the cloud* to cryptocurrency to the metaverse to AI and this might be the place where they run out of road because I don't know if the conditions for another bubble are quite right.
*Is the cloud a grift? I guess most people don't think so but it's definitely a way to turn you from an owner to a tenant and Amazon are very rich because of it.
Does top tier revolutionary OpenAI foundational leadership realise supremacy has one having to acknowledge, accept and engage opposition and/or competition in "embrace, extend, and exterminate” operations? That requires of one an extremely rare and raw mindset which one soon notices is absent and unavailable to one, and also certainly liable to prove one unworthy of future trust and generous grateful and gracious reward.
Is it that which is causing former helpers to re-evaluate their worth and jump overboard out of the frying pan ship and into lifeboats on fire taking hostile and frenemy fire ?
No, they are jumping out while they can still cash out, and into the friendly arms of suckers companies willing to hand them big signing bonuses, fully vested founder's stock, and golden parachutes when it all caves in around them. Same as every other high-tech vapo(u)rware scam over the last forty years or so.
... only to then dive, plunge, and hurl themselves off of the lifeboat on fire and into the flaming personal flotation device cupboard of forgotten transoceAInic history ... "all while taking hostile and frenemy fire" ... amidst cubist anthropomorphists !
Rolls right off the mawashi ...
Rolls right off the mawashi
More like a hatakikomi slapdown.. (or uwatenage if you must involve the mawashi).
Although it could be argued that the whole AI thing is more of a higi [1]
(For those not following, all the above are Sumo terms - the 6-a-year (ie every two months) bashos (championships) have highlights and some live action on NHK World. Best enjoyed with Roku or Ukiyo gin - or one of the better Suntory whiskies)
[1] Essentially - a move done by a wrestler that makes them lose the bout. 5 recognised higi types - all involve the wrestler losing without the opponent using an offensive move against them. AKA 'shooting onself in the foot' (metaphorically)
Altman had been the one who departed with the rest staying. But I guess that didn't actually work out. ;)
I shouldn't be surprised that in seven or eight years the only real benefit from AI/LLM will be that my (pen) tablet will be able to better read my handwriting. G-Board already does a pretty reasonable job and might currently use some basic LLL technology. Can only dream of the day when one might enter math formulae as one writes on paper (and have it converted to latex or mathcad etc. ;)
> Can only dream of the day when one might enter math formulae as one writes on paper (and have it converted to latex or mathcad etc.
I listened to a podcast last year about someone who came up with a way to do that but Apple cancelled his project. Their managers could not understand why it would be useful.
Well, it looks like Apple's IOS 18 let's you use their pencil on iPads to do some of that, as in: Step 4: Enter and Edit Math Problems (Handwriting) and also Scratch Math (and I would guess possibly many other underlying software for formula editing/conversion, solving, etc ... like Mathpix and what have you).
It would be great(er) though, imo, to have this (math software) run on a pad that actually feels like paper ...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
