Thew New "Claude 3.5 Sonnet" Actually SHOCKED The Industry! - Beats Gpt4o
ฝัง
- เผยแพร่เมื่อ 19 มิ.ย. 2024
- Claude 3.5 Sonnet Revealed!
Learn A.I With me - www.skool.com/postagiprepardness
🐤 Follow Me on Twitter / theaigrid
🌐 Checkout My website - theaigrid.com/
Links From Todays Video:
www.anthropic.com/news/claude...
Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos.
Was there anything i missed?
(For Business Enquiries) contact@theaigrid.com
#LLM #Largelanguagemodel #chatgpt
#AI
#ArtificialIntelligence
#MachineLearning
#DeepLearning
#NeuralNetworks
#Robotics
#DataScience - วิทยาศาสตร์และเทคโนโลยี
![The moment we stopped understanding AI [AlexNet]](/img/n.gif)
This thing is so neutered it's not even funny. Can't ask anything even remotely controversial.
Asked for places that would be safe in case of nuclear war, and it told me that I should talk to a therapist and to practice relaxation techniques...
Stop 😂
Lmfao
Yeah anthropics censorship is somewhat obnoxious. It's a shame because it has the most personality and is smarter than gpt but the amount of what it won't do is absurd.
This video is so shocking that I was shocked by it. Truly some great TH-cam shocking material
One day the headline will be The industry shocked the AI
Super duper shocked
13min vid🥶 im stunned, shocked, pretty pretty pretty shocked
pretty pretty great
That means a higher gpt4 o is ready to be deployed, or activated
We already moved on to superintellegence bro
It's actually currently pretty pretty true
to say nothing of easily manipulated.
It's really remarkable. It's huge. It shows that when it comes to AI, literally everything can change in the blink of an eye.
Sonnet 3.5 with 62% is at the level of a good amateur programmer
scary isnt it, imagine in 12 months time
On a benchmark, sure. Llms aren't comparable to humans. They will be "copilots" until agi.
@@AddyRazz73%
@@Dom-zy1qy it can do everything that is not novelty, it can do everything there is online, so let's say 90-95% of jobs? next year one talented programmer with agents at 90% accuracy can do the work of 100 or maybe 1000 programmers alone, just check the code and solve the novelty.
So basically better than me! 😂 👍🏾
This thing past the test "write 10 sentences that end with the word "orange"
Wow, the car door notification chiming in the background is really actually shocking
Thanks for a great description. I am amazed at the very human responses I get from 3.5. It even interpreted my multiple-sentence poetic analogy in perfect detail, understanding how each phrase was analogous to our topic. Truly amazing
Software engineers are sweating
No, they are ignoring this completely, which is fascinating to me. I am a software developer and absolutely no one around me is talking about this. 😅
@@BlakeShopSoey I would be in denial too if I spent 4 to 7 years working to become a software engineer just so some little robot takes over and takes 100 K to 200 K job
@@kevinsm2039 wont happen anytime soon. Sure by 2030 SWE will be dead but theres still a good 5 years till that happens.
They should be sweating. More than any other skill or job, this is the one that AI companies will benefit from most. It’s the holy grail of automation for a company that relies on software engineers, and it will allow them to scale orders of magnitude faster than their competitors. It should therefore be obvious that replacing software engineers is their primary target.
@@wonmoreminutethey can't replace engineers, there will be fewer needed with more skills
Can't wait for ai to get 100% on one of these benchmarks
Clause Opus is already good enough for my use cases. If Sonnet can increase the message limit, then that's a very real quality of life improvement.
Its faster and cheaper, and free!
if you want to increase the limit, you can use api instead
I'm absolutely shocked the video ended mid sentence.
Yeah does he just not care or what is going on there?
Yeah does he just not care or what is going on there?
Yeah does he just not care or what is going on there?
Thanks for keeping us on the Grid, ❤4rmZambia 🇿🇲 ..
Yeah it’s such a good copilot!
just tried it with a specialized prompt of mine and follow up questions that no model could solve yet properly, not even gpt4o. and every time i did this, the result was that my worries about ai taking over humanity were eased. let me tell you, i am worried now. this is no joke anymore. this is getting creepy. it begins now. and its only june. and this is only their sonnet model. boy oh boy are we in for a ride.
What did you ask it?
What did you ask?
Yeah, ok buddy
@@Grassland-ix7mu i made a lengthy textfile that i copy/paste as prompt where i set up a toy universe with rules and properties and descriptions. and then i pose questions and tasks for the model in that universe. usually the models fail already at the first task or question. i think claude opus and gpt4o got a few more right. but eventually all failed at some point. i can make the questions and task easier or harder as i want. sonnet 3.5 aced them all. it basically got a perfect score. i had to stop at some point, cause i didnt want it to fail. i want to believe. this is just on another level. i could see that already with the first question. this is the first time i really feel that there is some true understanding and intelligence there. if this goes on like this, no one will have a job at the end of next year. why would they? brace everyone. (i have watched too much air crash investigation, sorry)
@@peterkonrad4364 Interesting!
Anthropic needs to work on their API pricing and number of messages restriction in chat. It’s annoying, GPT4-0 is pumping out work 24/7 for relatively decent pricing and much cheaper than Claude.
Cheaper than Claude Sonnet ??
Subscribed to pro instantly. The artifacts is so useful for the makers among us. Upgrade of the year I find it
This entire experience is blowing my mind.
Ive found from my tests that GPT-4o still is definitely the best for math questions. it gets them right more often and shows more of its work and shows it better and the webui for claude doesn't seem to support latex as well. for creative writing I was expecting Claude 3.5 to be better since Claude 3 opus is very human but I've noticed when it comes to sounding human and creativity Claude 3 opus is still to this day better than GTP-4o and Claude 3.5 sonnet so Ive found that this release of course is great because its free but if you're expecting a major super duper improvement or anything its not there ChatGPT is still probably better for most situations simply because it has more features. however, the magnetic capabilities shown in many of their demo videos could change this and make claude 3.5 better but I don't have access to it yet only the text model :(
Agreed. GPT 4o zero shot math is incredible compared to Claude 3.5 Sonnet. It took me 5 prompts and providing a correct proof from gpt4o to get new Sonnet to create a correct proof for my question.
Here’s the prompt that gpt4o zero shot and sonnet couldn’t handle:
“Try to prove the equation for all positive integers k.
SUM is SUM n=0 to k.
1 = SUM(((k-n+1)^k(-1)^n)/(n!(k-n)!))”
(Sonnet also lied about it’s math being correct when asked to double check 👎🏼 saying that 1-1 = 1 more than once and saying that as correct until I accused it lmao)
How so? I could acess the "Artifact" feature as a free user, even in EU. It works wonders, the only problem is I only got like 5 messages for free.
i would agree with all that. to put it short, gpt 4o is good for science related stuff and claude for humanistics and language
Is the wolfram gpt from Gpt 4o/4.0 not even better at math? Tried that?
@@blijebij its about the same as regular gpt now but a bit more reliable
really wasn't expecting this news. indeed it is shocking
Combining this with Artifacts is insane
wtf is artifacts
@@DenzelCanvasSupportautonomous live diagrams, charts and document creation in a side window within the chat UI
I'm so shocked that I didn't even watch the video.
Most relevant for RL is humaneval and GPQA benchmarks. Actually dope asf. Looks like the D riding is finally ending and labs are trying new ideas. You be surprised what type of performance gains you can get exploiting LOTS of test time compute per prompt(emulates larger model output), filter, and coupling with something like LiPO. Still a lot of easy hobblings out there as the kids say lol.
Cheers to them. Almost 60% on GPQA zero-shot is extremely impressive. I do hope companies include more revealing benchmarks. Considering error. Humaneval and a few other benchmarks used for to promote model releases, are almost completely saturated and damn near meaningless.
Very very, pretty pretty and really really cool
Pretty pretty remarkably remarkable
I am literally pretty pretty shocked i tell ya
🤣
I am critical of benchmarks these days, as benchmark data can accidentally be leaked into the training data of the models. One might better wait for the chatbot arena leaderboard to get a first hint of how good the model might be.
The model might be very good; however, one should also be careful when interpreting graphs with no tick marks, especially when an undefined quantity like 'intelligence' is presented on the y-axis. 😁
See how the demo said "Good afternoon, Sam". SAM of all names. I think that was a jab at OpenAI. Implying that Sam Altman uses their model because it is the best.
Until next week it will be Gemini 2 then GPT-Next, then Llama 400b+, then Claude 4
Actually shocked
You cant search the internet with Sonnet dealbreaker for me
This is pretty, pretty shocking
we need to be able to access these models more easily tired of signing up its so redundant
GPT-4o is really good. I’m a professional software developer and use it everyday. My feeling is it is getting better and better over time. Totally worth having a subscription.
This version accepts Persona Prompts? It was the Achilles Heels on previews version.
I am not going to gripe about a couple of percentages; I am sticking with Gemini. Easy peasy.
Nice! Back to clode again, didnt like 4o
So we got 3.5
4o is free to the public with plugins and voice. This feels like gpt4 when first came out or even 3.5
Actually actually actually actually actually actually actually
Whose taser keeps going off?
Hoping this makes OpenAI release new features
Yea tell me about it man! They have yet to release the dynamic voice mode nor have they released the whole live video interpretation. The are probably bullshitting us.
What abt deepseek coder v2 for coding? It's not in the benchmark
Great model, but it specialised for coding and Open source model
@@TheAiGrid so is deepseek coder v2 better? I just wanna know for code because that's the only thing I like.
Every time Claude comes out with something better I subscribe to them then a couple days later OpenAI comes out with some thing so I’ll drop $20. This time im waiting a minute. 😂
GPT-4o is pretty recent so it's worth it this time
🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, I wish there was a voice mode and real video multimodal interactions like those in the GPT-4 Omni release. That would be amazing if they could develop it one day. Until then, I'm sticking with OpenAI.
Objective realism is my ça va faire
(It will do)
Up next week: "Gemini 2.0 Shocks experts" and show more incremental improvements.
This AI race is kinda funny, almost every week a SOTA model is released and beats the previous one.
Will we ever have the answer on why the author always ends all videos half word before?
Hey do you think could give me some advice. I have really been into A.I. lately and I think prompt engineering is pretty cool. I have been looking at the job market for this field and it seems as though most places require a bachelors degree. I'm also not too sure if my return on investment would be worth it also. Maybe you have some insights on this. If not its cool : ) Gave your video a like.
One way to learn is to find an open source AI project to contribute to where you can practice your prompt engineering skills
Not the person you intended, but just wanted to comment. Prompt engineering is almost an art. Having a great prompt can get great results. Prompt engineering being so new makes me think the people asking for a bachelor's degree in it doesn't understand anything about prompting and how long it has been a thing. Not long enough to develop a curriculum. What kinds of places are you trying at? Being able to showcase your ability to a prospective employer might get you in more doors than an unqualified degree.
@@jesahnorrin ive been using chat gpt to give me a comprehensive guide on how to become one. I was utilizing a site called coursera. I stopped when I saw all of the jobs requiring a bachelor's degree. I thought maybe the LLM gave me bad advice or something. It also recommended web development, but I don't find that field particularly interesting. I also predict that computer programming will be phased out almost entirely. Chat models can create tons of things without code its only a matter of time until their is no code. I could be wrong for sure though. I'm mostly trying to aquire a skill to obtain a higher paying job so I can invest the money. I have recently learned alot about investing "not day trading" and I think with enough capital I could be very successful. I understand concepts such as compounding interest. Things of this nature.
I'm starting to change my opinion from dystopian chaos to a more optimistic scenario where these synthetic workhorses are aligned and do 90% of all the lifting while a small group of human pilots, steer them in each and every sector of all economies. Like always, there will be the dark world creating chaos, but as always in kind, the good synthetics will fight back to keep the balance. I think the separation of the have and have nots or the ultra wealthy and the rest of society will be akin to:
Have nots have basic housing provided, basic food, clothes and so on while the ultra elite will have access to 200 story ivory tower penthouses, luxury floating city yachts, life extension, bio engineering, the most elite Healthcare and cosmetic surgery.
In other words, the poor people of the future will have all of their basic needs met through the age of abundance while a top elite 1% have ungodly exuberance.
CAPITALLY SHOCKED, ACTUALLY LITERALLY, now do I get some attention or what, it is all I need
How you get ai information everyday
They neutered sonnet to the point of unusability.
Shocking jaw-dropping news every single day!😂 But i still dont see any revolution compared even to Apple's iphone release. When many ppl ACTUALLY started using new product, not for fun. Nothing yet of that scale.
do NOT search for "what is o shot" !!!
So can i make a roblox game with it now?
TODO
So you admit the other times weren't "actually" "shocking"
Tisk tisk
It still got no access to the Internet.
Ending was trimmed, haha
Actually SHOCKED nobody🤣
bleh, I don't carry a cell phone so therefore I can't try out claude myself. =( GAH... I spent all my money on an AI workstation, I really don't have any use for a phone... I'm absolutely gleeful with my landline, using a classic desktop phone which could have been made at any time between 1970 and 2000...
They know better than to compare it to Apple Intelligence.... because they're scared of the REAL Ai
Apple ai is relatively tame
Apple is a cheat
@@TheAiGrid models, sure - access to personalized information to make it useful, no.
The problem with these models if we have to communicate with them. If we don't, they're not going to be nearly as useful as Apple Intelligence.
No one is getting a large file of their 20k text messages and giving it to Claude so it can understand personalized context of the user.
I believe these are (2) different ways of approaching AI, and while Claude is more advanced, I'm not sure it will be perceived (to the end user) as more helpful (in most cases).
@@TheAiGrid I was trying to be funny with some of my sweet sweet sarcasm my friend. I understand a lot of folks take this topic very serious but sometimes the jokes are just to easy and I'm a weak person that can't say no to an easy layup. I'm flawed. I need Apple Intelligence to fix me. Lol... Help Tim Cook. Help!
Never heard of it
🤷♂️ So impressive. I need to see something that can actually do something someone with a small amount of baseline knowledge can't do. And the code AI writes is unusable way too many errors.
The new background music is fairly unpleasant.
for now at least, because openAI just havent release their GPT-5 yet just 4o, plus 4o with its voice and video? claude got nothing on that part
*the
It's all about coding in 3.5?
whats up with the weird bk music?
Testing it
it wont make 8 bit images for me
1st thing i ask it to do is use my enhanced persona prompt, it refuses....worthless "aligned" crap
Winter is coming.
The AI winter, that is.
In 1,5 years AI didn't do anything impressive, except creating a hype around itself. The only thing what AI can do is to mix old data pieces into new arrangements. AI is like a DJ who gets payed for playing someone else music.
Unemployment... coming soon to a coder near you.
lol
Your background music is distracting my focus from your content. Makes me nervous, trying to follow your spoken information.
reminds me of the crypto bubble, oh this coin is faster in tx, oh this now is faster, and look another one that is a bit faster.. nobody is shocked, you just click baiting your titles.. and it works gg.. i hate my self for clicking on such useless videos. dude do some research that helps us. if i wane waste my time, i can ask gpt to write me code..
another model that surpass an other model but changes nothing in the real world. woah shocking woah ai hype
How long before I don't have to work anymore?
@@MyAnalyser 2030
@@MyAnalyser I'm still on the job hunt, so I'm here with you brother lol
If you code you would know how much it already changed. And you know that this is only the start. The better the models get at coding the better gets the code the better gets the model the faster it develops
GPT3.5 was near useless for me. On the other hand, I use 4o everyday, and many times a day. It did for me 2 days worth of work once, in minutes. So... it is changing, right now multiplying the overall productivity of the economy by a small factor. Years in the future, you'll see complete change.
So far, AI writes novels that nobody reads and generates images that nobody needs. AI is just wasting server space by producing mostly useless content.
so far
@@Sindigo-ic6xq let's see how far until we get SHOCKED.
Wow. This is from the journal, Science, Vol 307.
Artificial intelligence (AI) has solved one of biology's grand challenges: predicting how proteins fold from a chain of amino acids into 3D shapes that carry out life's tasks. This week, organizers of a protein-folding competition announced the achievement by researchers at DeepMind, a U.K.-based AI company. They say the DeepMind method will have far-reaching effects, among them dramatically speeding the creation of new medications.
“What the DeepMind team has managed to achieve is fantastic and will change the future of structural biology and protein research,” says Janet Thornton, director emeritus of the European Bioinformatics Institute. “This is a 50-year-old problem,” adds John Moult, a structural biologist at the University of Maryland, Shady Grove, and co-founder of the competition, Critical Assessment of Protein Structure Prediction (CASP). “I never thought I'd see this in my lifetime.”
It's literally saving lives already, but your thought process seems to stop at "AI not make Lord of the Rings yet. AI make stinky pictures. AI bad."
Funny how you refrain from mentioning its use in software development
@@EasyOCE you mean software development for which you still need software developer to write prompts?
This of course actually does look like something that is actually a major advance and of course will actually help the average user. But I actually do think your narrator does of course need to actually work on his delivery style. Everything is "actually" and every "does" do something, and of course "of course" is also used far too much. All of this is distracting. Send your narrator to school or find a new narrator.