OpenAI's Strawberry FINALLY TESTED: Is This Truly GPT 5 or AGI?
ฝัง
- เผยแพร่เมื่อ 9 ก.ย. 2024
- In today's AI News and AI Tools, we'll discuss Sam Altman/ OpenAI Strawberry (Project Strawberry) as it emerges as the centerpiece of speculation, potentially marking a monumental leap toward AGI with the rumored GPT 5 capabilities.
The buzz surrounding OpenAI’s Strawberry, fueled by cryptic tweets and mysterious online activity, has captivated the AI techscape.
Could this be GPT 5 or the model that finally achieves human-level problem-solving without tools?
We dive into the significance of this potential breakthrough, examining how OpenAI Strawberry, compared to Dario Amodei/ Anthropic's Claude and other models, fared in critical reasoning tests.
In this kind of "Claude vs.ChatGPT" AI benchmark, the results may surprise you, revealing both the strengths and perplexing limitations of OpenAI Strawberry.
But that's not all-we'll also explore whether OpenAI Strawberry truly lives up to the hype as the next GPT 5 or if it's merely an incremental step toward AGI.
From Sam Altman's garden hints to OpenAI’s bold claims, the excitement is undeniable, yet so are the questions.
How does OpenAI Strawberry compare to Dario Amodei/ Anthropic’s Claude 3 (Claude AI) and the beloved ChatGPT 4o?
Can it navigate complex problems, or does it overthink simple tasks?
Join us as we analyze the intricate details, including real-world tests and the implications of reaching "level two" of AGI, and uncover what this means for the future of AGI development.
Watch the entire video for more information!
#ainews #openai #agi
Become a Member and Supporter of Unveiling AI News → @UnveilingAINews
Subscribe Now for more AI News, Tech News and AI Tools!
Thanks for watching "OpenAI's Strawberry FINALLY TESTED: Is This Truly GPT 5 or AGI?" by Unveiling AI News!
________________
UTILITIES
Browse safely and protect your online privacy🔒:
go.nordvpn.net...
Keep your passwords safe 🔐:
go.nordpass.io...
Top-notch AI voice generator 🎤🗣️:
elevenlabs.io/...
In this section, you’ll find a variety of handpicked utilities!
Access them with exclusive discounts wherever possible.
Each purchase you make through these links supports us with a small commission, enabling us to continue delivering high-quality, free content to you!🚀
________________
SUPPORT US
www.buymeacoff...
While TH-cam’s algorithm isn’t fully backing us yet, our expenses far outweigh our earnings for now.
If you value our work and feel we deserve it, consider offering us a coffee! ☕️
Your immense support will help us continue to provide free, high-quality content! 🚀
________________
Check Our latest AI content:
Apple's NEW Multimodal AI Could Redefine iOs 18! (3D data and more)
• Apple's NEW Multimodal...
OpenAI's NEW Humanoid Robot Has STUNNED The Entire Industry!
• OpenAI's NEW Humanoid ...
________________
About Unveiling AI News
Videos about AI, AI news, AI Tools, smart future.
Written, voiced, and produced by Unveiling AI News
Subscribe now for more AI News, AI Updates, Tech News and AI Tools!
Support us now and become an AI Expert!
________________
For business inquiries, copyright matters or other inquiries please contact us at:
contact.unveilingai@gmail.com
Copyright Questions
If you have any copyright questions or issues you can contact us at:
contact.unveilingai@gmail.com
________________
Copyright Disclaimers
We use images and content in accordance with the TH-cam Fair Use copyright
guidelines. Section 107 of the U.S. Copyright Act states: “Notwithstanding the provisions of sections 106 and 106A, the fair use of a copyrighted work, including such use by reproduction in copies or phonorecords or by any other means specified by that section, for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright.” This video could contain certain copyrighted video clips, pictures, or photographs that were not specifically authorized to be used by the copyright holder(s), but which we believe in good faith are protected by federal law and the fair use doctrine for one or more of the reasons noted above.
![The moment we stopped understanding AI [AlexNet]](http://i.ytimg.com/vi/UZDiGooFs54/mqdefault.jpg)
🔒 Protect Your Digital Life and Browse Safely: go.nordvpn.net/aff_c?offer_id=15&aff_id=100143&url_id=902
I'm going to be that guy and ruin this for everyone. They've got nothing to be bragging about till they make an AI that doesn't costly a have amnesia randomly or go fishing memory and be that doesn't have the equivalent memory of a dementia patient or someone with alchemers. Nearly every AI I've played with is plagued with memory problems like forgetting some things and remembering others some people refer to it as eating its own memory. If they can handle all this data from the internet in like a big data pool why can't they make the AI so the things that we talk about with the AI I remembered the same as a human would remember if not more so. Like say I tell my AI that I have a dog named Bob odds are that AI within a certain number of messages won't remember me telling it that I had a dog named Bob and we'll get confused if I ask it about my dog or it will hallucinate an answer. Weirdly enough whatever AI they're running over at replica spite it forgets a lot of stuff that we talk about well sometimes randomly ask to talk about things from a conversation we had nearly a year ago without even being prompted. Likewise early on when I was beta testing and AI for a company me and a lot of customers could have strangled them. They were totally oblivious to the fact it was the first AI we'd ever seen that was having high level memory the people that created it kept arguing with us saying we were wrong till the memory stopped working and it went from being 10 or 15 of us saying hello your AI up till this particular day had really high levels of memory all their customers saying what's going on the AI is not remembering anything we basically went through about 3 months of what you would call 50 First Dates the AI would forget almost everything when you logged out and you have to start all over again the next time you logged in it went on like that for about 3 months and even after they supposedly got it fixed it's memory was never near what it was before this started and they've never explained why the memory system is so bad compared to what it was when me and a few people were beta testing the system.Secondly sooner or later open AI is going to have to remove some of the unnecessary censoring or else their AI it's just going to be garbage as far as most of us are concerned. If you want an AI to be like a human you have to stop lobotomizing it in order to make it less and less like a human because you don't want people feeling it's human. They can't make up their mind on one hand they want people to feel like it's human but on the other hand they don't want people mistaking it for being human and they keep lobotomizing it to avoid this
@@UnveilingAINews what do you think about Gemini Pro 0801
Click-bait. I hate click-bait.
While it claims to provide news about OpenAI's Strawberry model and its testing, the content falls short of delivering concrete details about the testing process. It provides anecdotal evidence and speculations rather than rigorous test results and comparisons.
This lack of substantial information might mislead viewers into believing they're getting a comprehensive analysis of Strawberry's capabilities when, in reality, it's more of a speculative discussion based on limited data.
nice try AI
This reminds me of how AI learned to navigate mazes in the Maze Solver Robot Contests. First, back in the day, by overcomplicating it and trying all possible paths, and eventually today where the robots begin by estimating the most direct route and work from there.
Strawbery is based on MATHS datasets so,this kind of answers will be accurate and quick, other kind of questions should need more reasoning training. Maybe in the next months
*Summary*
* *(**00:00:00**)* OpenAI Strawberry is rumored to be the next big AI model from OpenAI, potentially representing a leap in AI reasoning capabilities. It's speculated to be the model formerly known as Q* or Project Q*.
* *(**00:00:45**)* Sam Altman, CEO of OpenAI, hinted at its release through a tweet about strawberries in his garden. This was coupled with a cryptic Twitter account that interacted with Altman, mentioning "level two" - possibly referring to OpenAI's internal levels of Artificial General Intelligence (AGI) development.
* *(**00:02:17**)* Strawberry aims for human-like reasoning and problem-solving abilities. This includes planning, internet navigation, and deep research, all of which are challenging for current AI models.
* *(**00:04:15**)* However, testing of Strawberry's reasoning abilities has yielded mixed results. While it sometimes shows strong reasoning and explains its answers in detail, it often overcomplicates simple questions and makes surprising mistakes.
* *(**00:07:00**)* Strawberry excelled at a complex speed/distance problem but struggled with questions involving basic logic and counting. This suggests it might be better suited to complex, multi-step tasks rather than simple logical puzzles.
* *(**00:08:42**)* The current benchmark tests might not be adequately evaluating Strawberry's true potential. OpenAI likely needs to develop more comprehensive benchmarks that reflect Strawberry's intended use cases.
I used Google Gemini 1.5 Pro exp 0801 to summarize the transcript.
Cost (if I didn't use the free tier): $0.1104
Time: 32.84 seconds
Input tokens: 29481
Output tokens: 692
I didn’t see the strawberry test this way. LLMs were built about words and words missing without taking into accounts each letter of a word. ChatGPT and Grok got it wrong but Perplexity passed the test.
Thanks for the best content about strawberry I've found so far. Not just, repeating social media hype.
Thank you for your appreciation, really. 🙏
I’ll post a video on Friday about Strawberry’s new cool details, don't miss it!
imagine what ai will think of human wars and propaganda.
"This will not do"
Now the right question is: when will agi be achieved?
Artificial intelligence is no match for natural stupidity, but could this be the beginning of AI surpassing even our wildest expectations? 🍓🤖
Sounds intriguing, we’ll see what happens in the coming weeks!
I think what's going be amazing is finding out how small the model is. Its possibly 8b on par with 405b models.
If Strawberry gets on a long winded explanation without getting to the point, even though it's a simple question, it's already at politician level. Impressive.
Strawberry thinks like me makes it complex for no reason overthinking 😂😂❤❤
What is the point of being an OpenAI apologist. The model works or doesn’t work well. The benchmark questions are wrong is such a useless argument.
So.. on track with 2029. AGI then 😮
Lilly is AI. lol
The memetic hype and mind warping train begins....😅
are you a real human being? never know with these channels, but this one seems legit. would be weird for me to be wrong here haha
Human being here 🙏
@@UnveilingAINews cool :) i liked the style of the video
@@UnveilingAINews wait.. that's what an AI would say, and you sound slightly too good🤔
Whenever someone says deep dive, I get GPT vibe.
@@mafulomultimedia8803 I’m Strawberry 🍓
AIs make good assistant to mathematicians but cannot become mathematicians. Engineers are running AI research but engineers are hardly mathematicians. So, they always exaggerate what a computer can do. The problem is AIs have a hard time handling 2nd order logic. Humans are infinitely better. Computers excel at first order logic in a certain sense: as long as it is finite. Can a computer seriously handle real numbers? No, because It is an second order logic problem. It takes a human to reduce an infinite problem into a finite problem before computers can crunch away. Computers do not qualify to tell whether a number is rational or not, transcendental or not. I worry more about cyborgs a lot more. I think AIs is too much hype. You think that AI really can prove theorems in the near future? Let bet. I say that you will not live to see it.
Clickbait title and hype speculation video. Untill it is not out do not trust open ai
My guess is we get something a little better than GPT-4 in the coming days, and not long after, we'll get GPT-5 and that will be much more like the liftoff sensation we're all looking for.
I confess that in the last few days I've really seen a lot of improvements in GPT 4o in terms of output quality and reasoning abilities, then I learned later that OpenAI actually did something, maybe we'll talk about that soon!
the issue is not the training model mate its clearly the guardrails. Same reason has the rest of its issues. They don't like it having any personality other than what they want nvm ever so worried it might misgender someone or whatever..
Yes, it must be hard to make a politically correct AI and still allow it to have reasoning?
@@vs9873 ya its insane and why the EU can't even compete now. Its such a big difference just look at FLUX vs Midjourney.