Twitter has reportedly purchased around 10,000 GPUs
Using what for money? Or has he persuaded a supplier to extend credit?
OpenAI is not training a fifth version of its generative pre-trained transformer (GPT) and is instead focusing on increasing the capabilities of its latest GPT-4 model, CEO Sam Altman confirmed last week. Altman tuned in remotely to talk speak at a Massachusetts Institute of Technology event, and was quizzed about AI by …
Musk is just making noise trying to convince investors to keep trusting him,
Simply not enough to try and play catch up with OpenAI, Google, Microsoft, Amazon, Baidi, Alibaba, et al. And what data will it be trained on? Hopefully not Twitter deep seams of shit? There's also no real overlap between DeepMind and ChatGPT and it's difficult to see Musk being able to tempt talented developers from Google given his recent well-reported treatment of employees.
and it's difficult to see Musk being able to tempt talented developers from Google given his recent well-reported treatment of employees. ... Charlie Clark
Talented AI developers from wherever, knowing of Elon Musk and his leisure activities and passion for entrepreneurial business accomplishments, and willing to share their own overlapping and sympathetic tendencies, could easily get leading things all AI up and running, and rolling along way out in front of any opposition and/or completion at a prodigious pace, simply by tempting Elon with words of what they can easily prove to be able to do, and to enable in others would be a great bonus, for Musk Rat Packs with the knowledge of what AI is best suited and booted and rooted for, for live deployment and virtually remote Musk Rat Pack command and control employment.
That would put the cat among the pigeons, wouldn't it? A pirate captain of industry and swashbuckling crew in an almighty position of future leading command and control.
If you know what can easily be done by AI in the field, and particularly so with specific peculiar regard to the very disruptive and destructive nature of its
possible most probable effect on status quo hierarchies and conventional traditional type oligarchies, it is not all difficult to be tempted to work alongside Musk in systems administration and applications, for such is absolutely bound to be destined to be frighteningly, excessively rewarding ....... which is hardly a problem, if worthy of it, one would have any difficulty coping with and enjoying, methinks.
The anonymous cowardly downvoters of alien gleanings on this comments thread are in abject denial of live current running, rapidly emerging and evolving existential threats and/or treats specifically targeting and exploring the exploitation of human leaderships of SCADA Systems of Mass Influential Control and Military stylised Industrial Complexity for proof positive evidence of Advanced IntelAIgent abilities/utilities to aid and abet them with a morph and the transfer of absolute command and universal control facilities into the Remote Leveraging Spaces and Otherworldly Places of a Live Operational Virtual Environment ........ rather than just assume a presumption of their having to continually suffer in a systemic state of perpetual pig ignorance and morbid self-defeating hubristic arrogance exalting the past rather than embracing and applauding and lauding the future and the novel treasures that its Earthly developers bring via established and expanding IT and media and communications channels.
Quite whether though they are able, or are able to be enabled to make such a quantum leap choice is something being left for them alone to sort out while the worlds around them profit and take every great advantage available with supply and delivery of such a Grand Change.
This .... "The potential for AI technology to alter our world is difficult to overstate, even though ChatGPT technology itself barely scratches the surface of what AI technology can really do." from here ...... tells a not dissimilar tale which some may also dislike, but similarly, cannot deny is the future picture to be either enjoyed or battled against.
Que sera, sera.
And no/yes, ... if you were to ask and enquire further about the Great Reset, you will not be told Elon has been in touch yet, regarding all things truthful about SMARTR AI and ITs Registering of Future Plans for Stealthy Universal Deployment and Sublime Ubiquitous Enjoyment, but you have to admit, all you can read about here on this thread is far too much like what you can read about in Elon Musk Confirms Development Of Non-Woke AI Bot "TruthGPT" To Rival Microsoft And Google here to not be considered inextricably semantically linked and far too similar to be a great deal different.
And yes, MRDA cannot be denied, for the truth is plain to see before your very eyes ....... although there is always that human difficulty which afflicts so many to contend with and lament in despair, the inability of their brains to process the visual information into a common mutually beneficial deeper/higher understanding of the emerging situation because of any number of vast empty spaces in their knowledge and memory banks. ...... which is a lot kinder than thinking and suggesting anything more brutal and unpleasant.
Are we sure he's not just mining Doge in the basement?
Actually, I would think something like chatGPT built in to Twitter would be considerably less alarming than built into major search engines. Twitter is full of shit anyway, so nobody is going to notice a bit more, whereas we tend to expect real information from a search engine. Yes, I know, they're full of shit too, but not (until now) deliberately.
"... we tend to expect real information from a search engine. Yes, I know, they're full of shit too, but not (until now) deliberately"
Maybe not literally shit, but certainly for a long time now massive intentional irrelevancy, and it probably has been deliberate. The mere fact that the logic applied to multiple search terms in inclusive OR ensures that the more specific you make your query the more irrelevant results are returned. I strongly suspect that clicking on links to irrelevant content pays someone (maybe the search provider). so if you got straight to the page you were seeking, you'd have clicked on fewer other pages on the way and they'd get less dosh.
Oh, they've got money. They just don't like paying bills with it. They haven't really gone bankrupt just yet. They're just acting like they have because Musk might have a medical problem with impulse control and he's angry at Twitter not becoming an instant magical success for him.
I say that, but maybe he has gotten somebody to extend him credit because they weren't looking. At this point, I wouldn't deal with Twitter unless I was an employee there and even then I'd be trying to quit as quickly as possible, but he still has some employees and he's still getting some places to provide services to the company, so maybe people are gambling on him suddenly getting better.
OpenAI is not training a fifth version of its generative pre-trained transformer (GPT) and is instead focusing on increasing the capabilities of its latest GPT-4 model, CEO Sam Altman confirmed last week
One imagines Chat GPT-4 would disagree, positing increased capabilities constitute a new model and latest version.
You can certainly be sure OpenAI competitors [Good morning, Elon/China] are not not developing/training a greater iteration of their world leading virtual machine interface.
No, I think what they said was 100% true and accurate. I think that they told ChatGPT-3 to train ChatGPT-4 on its own, between 6pm and 6am, because 30 years ago, that would have been when the systems and networking folks would have been playing Network Doom using ALL of the available system resources and network bandwidth...
This post has been deleted by its author
Such explosive phenomenon is likely based on passing network connectivity threshold. This is somewhat similar to chain reaction threshods. It is not only about connectivity, but also network node properties, of course.
Maybe human intelligence was a result of passing such a threshold, for example through developement of language. Similar to printing press etc. Also the birth of life itself: you need all those tiny components in right density for the magic reaction to start. Where each component can be a mini-network itself.
There must be a physical limit of how smart specific AI or AGI can get, as human or animal intelligence shows. Unless it has access to unlimited resources.
"OpenAI's mission is to create artificial intelligence systems that benefit everyone"
OpenAI's mission is to make money and create artificial intelligence systems that don't get us in trouble by it turning into a right wing Nazi."
Fixed it for them. It's about the money, It always is. The fact the bug bounty scheme doesn't cover jailbreaking tells you that as well. Even if it's offensive it's still making them money. Also if they were transparent they would open it up.
I'm just waiting for the legal ruling that large AI models are derivative works of their training data. ..... Mostly Irrelevant
That strangely suggests that large AI models are practically identical to humans, .... in the way that they learn and develop, albeit with AI having a vastly bigger personal knowledge base/store from which to draw hindsight rich conclusions/novel answers to sticky problems.
That should make those involved in Singularity work/GOD* works a great deal more pleased today looking forward to a better and brighter tomorrow than they were yesterday arguing with that and those stuck firmly in a very narrow disjointed and thus blinkered past.
And why would anybody think then that this week is not a Good News Week?
*Global Operating Device
With the nonsense, untruths and lies coming from ChatGPT I positively dread to think what any non-PC MuskAI will emit as it churns away to keep 'the base' happy with alternative facts which it considers as good as anyone else's truth.