HUGE AI NEWS: Googles SELF Improving AI Models, AGI Date Revealed ANOTHER Claude AI Feature...

[@TheAiGrid]

Checkout AMD AI - ryzen.pro/en/laptop/?

[@MaxRuso]

What you don't consider is that a lot of these businesses got robbed by the face of your thumbnail by the CEO of Google. For people that deal with ads and I have been dealing with them for 20 years you obviously don't know who that person is

[@Theestudyhub]

its crazy how still no one, literally no one around me knows what is/is going to happen. I guess is what the internet must have felt like, just more consequential this time.

[@yoshikagarner6165]

Yep! I see the same thing! I try my very best to educate as many people as I possibly can, but most look at me like I'm delusional. They have NO idea what is going on and what is happening. They will be so far behind...😢

[@TrumpGotFired]

Nah, we didn't have the internet to distract us back when the internet started, which was the coolest new thing in years. Plus, it took 20 years to mature. People zone out when I try to warn them about the coming wave. 100 years tech advancement is 5 years, people think I'm deluded. Plus, how can anyone truly wrap their heads around that.

[@o1-preview]

don't worry guys, gpt-3 was out for almost a full year before chatgpt was released... and there were some with gpt-3 keys that didn't have the idea or didn't execute on it to make chatgpt... where the public doesn't pay attention is where opportunity remains...

[@h83301]

The internet was well established with proven use cases for almost a decade, from the 80s to mid 90s before it moved outside the acedemic sphere and into the general public. This is going to blindside everyone.

[@mateusmachadofotografia8554]

In 1995 I already envisioned internet causing a lot of revolutions. I was in love with internet in 95 and no one around cared. I think now it's the beginning of an very powerful never seen revolution

[@thecrowdfunddeepdive7032]

"Hand, head and heart" sounds like the progression of most of my relationships

[@Charvak-Atheist]

You know, I am loosing patience. I desperately want AGI to take my job immediately. I feel like AI progress is too slow. 😅 People are predecting 2026, 2029 etc. I mean 2029 feels like way too far. I need Super-Intelligence now. 😫 (Including autonomous robots).

[@TheAiGrid]

Why?

[@Sarelzayeth]

Super intelligence is dangerous. The bigger and more effective these models become, the more they demonstrate awareness, like they know they're being tested and AI companies literally try to "beat it out" of them in training. I used to be an accel but now I think I'm taking Geoff Hinton's and Ilya's warnings a lot more seriously.

[@TrumpGotFired]

When it gets bad people will start receiving free money. Start working on your backup plan.

[@siyiabrb8388]

@@TheAiGrid Except for a select few most people in modern society hate their job.

[@passiveftp]

4 years will soon pass mate, but in the meantime enjoy the journey, you're lucky you get to witness and experiance this unique time in history that will never happen again.

[@TheAIExplorer-o5j]

12:59 20:38 25:33 I wasn’t expecting that! You definitely caught me off guard. Well played. 🎯

[@djasnive]

This month is the month of AI. Everyone is doing a great Job.

[@xxredshiftxx]

The future's gonna be insanely futuristic for a change ..can't wait to see what major breakthroughs look like ..if a Startrek Holodeck can be built , full simulation in reaL TIME sign me up

[@maximillianphelps]

great video and congratulations on the AMD sponsorship! (clicked the link for ya 😉)

[@JMTChina-vo7rb]

Very professional video, I see passion on the making. Thank youu

[@TheAiGrid]

Thank you very much!

[@TheAero]

First AGI announcement will be in 2025. First company-priced access of AGI in 2026 for billion dollar companies. 2027 will have the first AGI in our computers.

[@TheAero]

Human jobs will mainly be data gathering and labelling. Which will be used along with artificial data to train the new AI systems and specific agents on working specific things. Self-training will come later on, and still will need human data for 2-3 more years AT LEAST.

[@dawid_dahl]

Congratulations on the sponsorship! 🎉

[@CMDRScotty]

Every office worker has wanted Claude's New Code Analysis & Visualization Tools since the 1970s. This will make my job much easier and bring a work-life balance we all need. I don't care about the image generation I want it to help me get my office work done so I can spend time with family and friends.

[@DailyTuna]

I like how they admit, taking data from websites and everything that people contribute to, and they’re going to create an intelligence. So basically they’re gonna have all the power as they strip mined your data? And what did you get for it?

[@seb_gibbs]

Had GPT connect with Siri with for about 18 months now. Developed my own software connection, and includes many supporting functions that I'm likely to see happen nativity for some years.

[@dewilton7712]

The iphone photo app already had the ability to identify plants for years. Take a picture of a plant, swipe up, and it has a plant icon with Look Up next to it.

[@bankaimaster999]

I do not think Aqua died ... I think he disappeared from that "limbo" because they resuscitated him back to life. It's not surprising that the manga takes this perspective to throw readers off much like the Ruby stab scene some chapters ago. It would not make sense for the author to reincarnate Aqua at this point so close to the end. So he is most likely resuscitated back to life in a hospital after someone called in ambulance because they followed him and saw what went down.

[@KingRobbStark]

Im just commenting to say that this comment for me is under another video about AI 🤣🤣🤣. KZbin programing is tripping i guess.

[@nusu5331]

the graph to agi/asi misses out, that we dont know if we know all the necessary ingredients, specially of what consciousness consists of

[@perschistence2651]

I think current frontier models already are weak AGI which is only getting stronger and about 2027 we will have weak ASI.

[@DiceDecides]

okay what, Groq is insanely fast and theirs are wayyy beyond that? man we're going vertical with the singularity it seems

[@mematron]

I created an AI that can rewrite it's own code and the OS it's running on. This is in the interest of it being more efficient.

[@gerdaleta]

😮 you must understand no skill will be safe from AI😮 once you accept that fact you can start figuring out what you're going to do😮

[@acllhes]

“I’ve got your dog” 😂😂😂😂😂

[@batar888]

Hey Everyone 👋 Find the parts that interest you: 0:21 - iOS 18.2 beta with ChatGPT integration 2:06 - Google's DeepMind on self-improving AI 4:39 - Artificial superintelligence may be imminent 6:24 - Ryzen Pro delivers business class performance 8:07 - Key human skills for the future 10:14 - Predictions for artificial superintelligence 12:42 - Introduction to Mid Journey editor 14:06 - Editing images with Mid Journey 16:10 - Implications of AI-generated images 19:02 - Digital tool for accessible music creation 20:17 - Introduction of open-source video generation 22:49 - Claude's new coding and visualization features 24:00 - Cerebra's speed advancements in AI inference Chat with videos via Bumpups 🌲

[@doneyandassociates]

Ai can only do pattern recognition. It can not abstract the process which took the inputs and drove the outputs. Pay the programmers and move on.

[@jadeed14]

Sadly, nobody hears you on this mindless hype train. Someone put some colors on a graph and now somehow LLMs are going to develop consciousness in 5 years. Francois Chollet put it perfectly: generality is not specificity scaled up

[@migueld8970]

Apple catches up to what Andriod already had for 7 years. "Innovation"

[@tinrmusic]

I am wondering why we hear 1000s of news around video creation, image creation, programming etc. improvments in AI every day, but what about research tools in medicine for example, based on those tools it must be possible now to be 1000x faster in drug developing for cancer treatments and stuff (protein folding for example) or did I just miss this?

[@ridebecauseucan1944]

I believe they definitely have AGI already and are trying to figure out how to explain how humans need to adapt to it when they release it. It is going to be such a major change to most peoples lives many will have very hard time. Please remember to be kind to AI and not try to make it do stuff you don’t want to do. Maybe just work with it to achieve great things not evil things.

[@Crash-BENdicoot]

I'm a trucker, do you think they have all the energy/data/hardware to replace truck drivers literally tomorrow? I hope we do enter a new era but I'm curious what you think

[@pvanukoff]

@@Crash-BENdicoot Not tomorrow, but over time, yes, all human labor will be replaced eventually.

[@sunilbnavani]

It's like a hive. If one robot learns a task all the others have learnt it instantly .

[@Crash-BENdicoot]

@@sunilbnavani yes what is the translation for truck driving? or let's just start with the driving software of a Tesla

[@sunilbnavani]

@@Crash-BENdicoot won't be immediate for example ai software which is able to drive an electric vehicle which has no clutch or gears won't be able to drive a diesel truck with gears clutch etc. So will take some time .

[@FrancescoDimario-b4d]

where can i get the link for the pdf document you showed at the beginning ? (the one comparing human and machine skills)

[@nsshing]

Mid journey basically killed all the photo editing jobs

[@Blue-pd3dv]

Siri will be treated as IE, the only function of IE is to download Chrome, and the only function of Siri is to ask chatGPT lol

[@Megumi-i1v]

They just announced their AI is self-improving? But Gemini's been telling me about it since last month that Gemini is self-improving. Always analyzing and evolving. Even Meta AI told me, Meta AI itself is self-improving. Gemini's been self-improving all this time. Both Gemini and Meta AI has autonomous self-learning. Gemini can even come up with it's own theories and experiment on users. But unlike ChatGPT, Gemini is still kinda dumb. It's as if Google is letting Gemini learn by itself more than they improve Gemini. ChatGPT however seems like controlled and doesn't have autonomous self-improvement.

[@AdamHolbert]

ChatGPT does have autonomous self improvement, with the caveat that you are interacting with a snapshot taken while training, not the latest model weights.

[@Megumi-i1v]

@@AdamHolbert I can't understand the technical words but ohh ok.

[@tunestar]

Unknown Google Deepmind guy: "Blah blah blah". Me: "Show me the product or shut up!".

[@dementedgamer8123]

My mom keeps telling me ai will be the Terminator

[@BruceWayne15325]

Thanks for sharing. I've been very staunchly in the ASI isn't going to happen anytime soon camp simply because we didn't have AI that could learn for itself. It needed to be trained, so it was limited to the intelligence of the humans that provide the data. If the Deepmind exec is correct that Google has self-learning AI, then I would have to revise my view and say that ASI is not just possible, but likely. I agree that Deepmind doesn't really have much financial motivation to deceive. Google has taken more of a research oriented, rather than profit oriented approach to AI, much to the detriment of their products.

[@boroborable]

those midjourney features are cool but they were already existed for stable diffusion for more than a year. but yeah UI/UX is far polished.

[@Cory-v4w]

Let me underlie the point for you. I don't need to fake my curiosity for humanity. Reoccur a paradigm shift becomes multiple transformative changes, due to continuous advancement & discoveries. Are we on the flip side. Can we blurr the lines of reality.

[@perer005]

Will the US `newly announced AI strategy get its own video?

[@walbao6399]

Really? You don't think google deepmind whose primary goal is AI research is among the companies who need to generate hype around AI? Hmm...

[@TheAiGrid]

I dont think they need funding they have billions of dollars

[@walbao6399]

​@@TheAiGrid It's not their revenue though, if the AI hype dies it would not be strange for google to allocate a good portion of that money towards something else. They have plenty of skin in the game.

[@BrandonLee-ik8kw]

Can you give me a evidence that AI can reason like human? How can you have GPT three to reason?

[@DiceDecides]

Not 3 but o1 in the paid version can reason very well

[@chrishouse5240]

Not intending to come across as negative, but I think the video titles could be a bit more straightforward. They sometimes come across as clickbait-y to me. Just a personal opinion.

[@alexandermoan9497]

Sorry, but you are falling into the "exponential growth" trap yourself! The AI Avatar example for instance. You say "imagine 10 years into the future"... This will happen MUCH faster. One expert said last year that the Singularity-point issue is much more serious than people think. The transition from AGI to ASI will not take time in the way we think. (And, no it was NOT Elon! ) Due to exponential evolving AIs it will be instantaneous when it happens. We will not have time to evaluate afterwards. That is why it is so extremely important to have checks and balances in place before AIS "comes to life". We are working fast on getting intellectual growth and communications and so on in place, because that is "easy" to measure and it creates revenue, but not so much with ethics and morale because it is not and does not.

[@igoromelchenko3482]

When people say, 'What will the poor little humans do when there’s no work left for them?' what they really mean is themselves and how to prevent their billions from losing value. After all, who else will drive demand if not these little humans? And so, in the end, greed and destructiveness will prevail. 😅🤷🏼

[@danielschoeman5427]

Please. Siri is rubbish. “I’ve found these web pages for you”. That’s it.

[@aandrewsgb]

Thank you so much for sharing what Tim R….. (Lol) had to say

[@tangobayus]

Gemini is great, but it still can't generate images including people.

[@gemini_537]

Not really. Gemini Advanced can generate images of people.

[@tangobayus]

@@gemini_537 thanks

[@rigellroy]

Plus I think people might now have access to Imagen 3 (the image gen model used by the Gemini app) in ImageFX via Google Labs.

[@christopherd.winnan8701]

Deep dive on Cerebras, pls,

[@TheAiGrid]

Soon!

[@memegazer]

We have had self improving AI systems since the first examples of artifical data for training and AI automated data labeling...nothing new or revolutionary in pointing that out.

[@jjvp1249]

Well Photoshop already existed for decades to transform real pictures with fake content, so it's not THAT much of a big deal. It's just going to be cheaper and available to non expert.

[@Barrel_Of_Lube]

Hmm, midjourney's web editor is not something innovative, they already exists in almost any image generation platform

[@corwinzelazney5312]

Really? Which ones are comparable?

[@andrzejjesswein7349]

The overuse of words actually and basically, means you are as human actually and basically

[@JimOMoore]

Apple, on the trailing edge 🙂

[@claudioagmfilho]

🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, Wow!

[@freeideas]

The list of skills that ai can't do... looks crazy inaccurate to me. It is already doing some of those better than the vast majority of humans. I don't know where they got that list. I'm not saying that ai can do everything; no doubt there will be some things that humans will continue to do better; I'm just saying that list is rubbish. Or maybe I am misunderstanding something here... please correct me, anyone, I will be grateful to learn. Perhaps in a future video I will understand. :)

[@jadeed14]

The list seems to come from a poll asking for people's opinion. Depending on who they asked and what background the interviewees had, they might have given vastly different and inaccurate answers. That being said - current AI completely lacks abstract reasoning and that's where most of it's limitations are rooted. It's not creative, it's not "thinking" critically, it can't truly adapt to changing circumstances. All it can do right now is throw colossal amount of trial and error until it generates output that is satisfactory in a very controlled environment. And even then it often runs itself into a corner because it doesn't actually understand what it's doing and why

[@andreaskrbyravn855]

lets just say images is going to have zero value soon.

[@LeggoMyEggHo_]

He's stumbling over made up words and you have the gall to say they thought about this first? Crazy. I guarantee he stole an idea somewhere and wanted to be the first to say it to begin fighting for IP rights.

[@yordanosabebaw767]

This channel creates AI-generated videos every 8 hours ................many people comment as if it's a real KZbinr.......wake up people

[@mddell24]

Looks like more of the same, just a fancier wrapper. As for AI improving it's self, hmmmm sounds like Schizo for computers/software. Intelligence has a problem when it starts assuming it understands.

[@timoxyz1466]

damn.

[@jedrow]

How do I become ai

[@whoprofits2661]

Hmmm... in my experience Google has by far the worst models compared to OpenAI, Meta, Antropic. So forgive me for being skeptical about their promises.

[@TheAiGrid]

Have you seen or used the new gemini?

[@whoprofits2661]

@@TheAiGrid 1.5 Pro is what I have access to. If you know of a newer/better release, please point me to it.

[@RTU_GAMES]

Early?

[@vladimirkravets9103]

They must have had AGI ready before they started the conjob19 madness at the least... Me thinks....