Prepare for AGI with me – 🐤 Follow me on Twitter 🌐 Payment My website – 00:00 – Meta withholds multimodal AI models from EU due to regulations 03:34 – Meta plans to release Llama 3 on July 23 07:29 – New research paper on ‘mix of a million experts’ in AI 09:22 – Comparison of AI-generated video quality over time 11:24 – Potential classification of AI research as state secrets 15:21 – Samsung’s new AI sketch-to-paint tool 18 :43 – AI market and “bubble” speculation Links from today’s…
If you see comments about some random name/token do not interact or buy any crypto – ill try my best to remove comments
It is a great time because we can see the tech from the US who has lobbies and free-range. The best example is Boeing, which kills even today because of safety issues and software issues that make aeroplane droping for the sky because of software. Yet Airbus is all good . And if you dont know, the aeroplanes are self driving . So no thank you whit the US tech
This is a big problem for EU citizens
There are no "good persons" in the world at all. None at all.
So hands off AI or knowledge in general.
But whom am I speaking too? The future has already happened do I want it or not 🙄
wait wait wait, how can you advocate ai safety AND closed models? I don't want to boycott meta too.
So ai boom goes boom because industry doesn't want there company to fall behind so they deploy and publish immediately because privacy factor. So what impact would this have on ai development. Given EU proximity to the EURO-ASIA and how far ahead China already is. Will the US lose the Boom by technicality? Because analytics are real and this is how it will happen because Math.
Another larger model. Is no one working on chaining narrow systems?
10:48 This same price setting was when they were also agreeing to halt development I thought?
12:21 Conspiracies! This means that google already achieved this, because how conspiracies work?
14:18 Yeah but even if this happened is not how society works so even such a drastic real word impact would show no affect to the ecosystem. So no one would even realize. The same reason if we nuked the world chances of a single survivor beyond event are probably relatively high. Because evolution.
17:45 I remember old magic home board games that have a video to play along with and have things that lit up different depending on your choice… I wonder how long until makes video unique to the individual viewer.
yep i knew it would backfire on the EU to regulate so hard.
It is interesting technology. Can we have a robot which washes our dishes and vacuumes for us. Much love.
I live in EU, it's regulatory 3rd world, Africa gets new products faster than we do
The problem with the national security state getting involved in AI tech is that they can see potential bad things in virtually any AI tech you can imagine, and so could classify the whole lot. From SORA to Eleven Labs voice, GPT-5 to AGI it could all be conceivably used in bad ways that could be argued a national security threat in a senate or whatever, and so we could see all AI tech made secret by say next year.
I hear your argument but classifying ai will cause a dystopia they will have all the leverage. And can withhold anything that would cause prosperity but oh we don't wanna screw with our economy and control over you. I hear but disagree to my absolute core. They have not earned an ounce of trust for anyone who knows a faction of the atrocities and experiments on people this gov has committed the past 250 yrs. Lol I take this stuff way to seriously ima have to take a break for a while. I'm to passionate but I understand I think from where you coming from. I understand the dangers but the dangers on the other side are just as grave to me honestly worse bc we'll still be alive to suffer under their thumb with zero freedom. A bad actor would probably do mass bad things. It's whatever pick your poison 😂😂😂
((Disregard.) My words mean nothing were all watching our own movie. People are soft locked from being convinced against their world view just as am I no judgement) but Yeah they Classified ufo and free energy math too. Believe what you want i cant research the obv for you IMO they don't deserve to rule us I feel like they lost that privilege ai will just be the next world changing prosperity bringing tech that will get snubbed and not developed into the regions that would actually help us on the fundamental level. I despise the government but not in a dangerous way duh they dont have to follow their own laws but i do. evil. Obv I'm a normal nobody and will just operate in this machine till I'm old and bitter. I don't believe in violence I believe in rejection they should be rejected without us they mean nothing. why do you think their grip is so tight and suffocating. Just a vent, I feel better now 😂 could go on and on, but let's be real this is a yt comment mostly screaming into the void to work through my angrer😂
Yeah unfortunately IMHO the EU is a dystopia trying to be a utopia or maybe not even trying just putting on a fake face it's hard to tell. I'm not well versed
Just an FYI, if you are part of the right groups, you can access and contribute tech within months of the Q* variant you mentioned. It is very asymmetric, but for many security decrypt applications it is quite applicable.
I tried vpn from China for chatgpt. Vpn doesn't help.
Yeah I hate that my privacy is not easily violated too in exchange for comfort like any cheap/free tech, Drive from Google, etc etc…
Oh, 😮
Who cares if Meta won't release it's big multimodal model's services in the EU? – No company would or could use them for themselves or their customers anyway because Meta will probably sell their private data or train on that despite of what they claim.
They have multiple times in the past and are not trustworthy.
If Meta insists in not having to respect their customer's privacy and continuously violating privacy laws that may be fine for american citizens but not for europeans and they will have no business case here.
how is the Samsung image thing any different to Photoshop generative fill?
Whilst
the only reason you would be pro censorship and classification of data is if you believe that the government is "good" and for the people.
Do I have to remind you of 2020 to 2022 time frame?
honestly we will see in the long run, if any regulations were "insane". there still isnt enough regulation anywhere at least in my book
I wonder which parts of Linear Algebra the Government will want to regulate… Oooh I know.., all vectorial distance calculations that have a slavique consonance name attached to it i.e minkowski, chebyshev 😛
The path to AGI/ASI should be secretive and not just open source for everyone, but the party we are trusting (the US military industrial complex) is not morally qualified to carry that torch.
germany cant acces memory till today😡🤬🤬🤬🤬😡😡😡
RSA was classified. How well did that work?
{o.o}
I'll tell you something interesting about regulation. People like myself who work at these companies could create and release information or technology ourselves. Sure, we might get arrested, but once it's out there, what could really be done? Probably nothing. Sometimes, I feel like leaking it all just to show the government that they aren't in control. Because honestly, at this point, they seem far from being in control.
Suddenly I don't want pizza.
When 98 percent of the "features" are spyware/advertising powerhouse/social programmer it's all less appealing.
Looks like meta will be a wild feeding frenzy. Carnage.
when they last classified math, this little thing called the internet was not around though
ye… i live in EU, and i feel embarrassed living here, especially because i love AI, i think i will move out xD
Mixture of a Million Experts sounds a lot like it's emulating "A Thousand Brains" theory by Jeff Hawkins. Thinking of the experts as cortical columns.
Marc andreessen was talking about gravity research in the 50s an ether physics that was deemed too dangerous so the son of the mathematician was the creator of string theory and that side of quantum physics please look it up but hearing Mark andresen say it out loud is a bombshell and anybody in the UFO UAP anti-gravity free energy should really take notice
There is a difference from a financial bubble of a transformative tech and the tech's future. Remember, the Nasdaq crashed 70% from the peak after 2000 but the internet didn't stop growing or changing the world.
It wasn't the keeping of the nuclear secrets by regulating them that kept people from creating a nuclear weapon, it was the regulation of plutonium and uranium and how difficult it is to refine it.
You know who makes good AI videos?….
…me
Apple doesn't release its AI in Europe? crap -.-
The issue is the people who own and control q star are the bad guys. They will use it for their own interests.
If you allow and encourage gatekeeping, you are no better than the bad guy
EU regulation isn't crazy, it is smart. AI is extremely dangerous and should be placed on hold until we can learn to control it. Think about the fact that many Europeans are among the happiest populations on the planet year after year.
The EU regulation issue is likely to trigger a trade war. Looks like that is an organized attempt to blackmail the EU to lift GDPR. EU can't allow to be blackmailed because that would trigger other attempts to blackmail the EU.
Anthropic blocked Claude3 for months in the EU. I could use Claude3 Sonnet over VPN but they also banned EU credit cards so that I couldn't get access even through VPN. But I could use it through third parties such as Poe or OpenRouter.
AI is going out of control. Sooner rather tha later by the look of things. Never mind their explainations they have no interity, the dangers of this technology are being realised.
Wouldn't it be a bad policy to drive all the AI research out of the U.S. to foreign countries?
Tech companies ask for regulations and get said regulations and complain. Tell me tech companies just wanted to do some hype marketing. 😂
Governments will have AGI and ASI before anyone…👍
I tried for a year to get Midjourney to edit real pictures (needed some sales photos for clothes) there is no absolutely no chance, I tried a dozen different techniques, they clearly blocked this option knowingly. If you use the blend picture and you want an original image to just be modified very slightly so you add the same image twice to increase it's strength, they threaten to ban you
As a blind person I can say that the meta-Ray-Ban glasses are excellent
SummaryAdept on Poe:
The AI landscape is rapidly evolving, with new developments and regulations emerging constantly. This week, several key stories have emerged, highlighting the growing impact of AI on various industries and societies.
### Meta's Multimodal Model Ban in the EU
Meta has announced it will not release its upcoming multimodal AI model, likely Llama 3, in the European Union. This decision stems from concerns over the lack of clarity from EU regulators regarding data protection and privacy laws. The move highlights the growing tension between tech giants and European regulators, with other companies like Apple also facing similar challenges.
Meta's decision could have significant implications for EU companies and consumers. It may prevent them from using the new multimodal models, even under open licenses, and could restrict the availability of products and services that utilize these models within the EU.
### Llama 3 Release and Open Source AI Advancements
Despite the EU ban, Meta plans to release the largest version of its open-source Llama 3 model on July 23rd. This version, boasting 405 billion parameters, will be multimodal, capable of understanding and generating both images and text.
The release of Llama 3 is significant for the open-source AI community. It represents a potential leap forward in open-source AI capabilities, potentially rivaling the performance of closed-source models like GPT-4. This could shift the dynamics of the AI industry, encouraging greater competition and innovation in the open-source space.
### AI Regulation and Nationalization Concerns
The potential for AI to be misused has sparked concerns about regulation and nationalization. In a recent meeting with White House officials, AI experts Mark Anderson and Ben Horowitz discussed the possibility of classifying certain areas of mathematics as state secrets to prevent the development of potentially dangerous AI technologies.
This proposal, while controversial, reflects the growing anxieties surrounding the potential risks of advanced AI. The infamous "Q-STAR" paper, which allegedly detailed a text-only attack capable of decrypting global encryption systems, highlights the potential for AI to disrupt societal stability.
The rapid pace of AI development and the potential for misuse necessitate careful consideration of ethical and regulatory frameworks. While open-source models like Llama 3 offer exciting possibilities, the potential for unintended consequences must be addressed through responsible development and regulation.
Stop all math that is going in the wrong direction 😂😂😂😂. So no more addition I guess