@mrencyclopedia @freevolt24 @arstechnica
The frontier models (#o3pro) is so hungry for context that it will pull it from adjacent areas and the end user may mistake it for hallucination.
TLDR; The newer models are so good, Apes can't tell if it's hallucination or if they too dumb for the machine 😬
Looks like OpenAI's latest gizmo, #o3pro, is just another way #for them to say "we're #hungry for your money" 🙄💸. Cue the #bloggers who were first to #review it, acting like they've discovered the next sliced bread when it's just a #toaster in disguise. 🍞🔥
https://www.latent.space/p/o3-pro #OpenAI #critics #money #HackerNews #ngated
First thoughts on o3 pro
https://www.latent.space/p/o3-pro
#HackerNews #FirstThoughts #o3Pro #TechReview #LatentSpace #Innovation
Here is a mind-blowing #o3pro prompt for you to try on the brand new #OpenAi model;
"Based on what you know about me, what would make me 50% happier"
- via Nate B. Jones
This new model is another step towards #AGI , chances are your prompts will be wrong.
You don't think about it as a Chatbot but more as a deep context strategic adviser. It struggles to "dumb down" results to your level.
In my case affording access to the model ($200/pm) would make me 50% happier. Simple minds 😁
According to Jones, this is hands down the best #LLM #AI model on the planet right now.
(This model is so new the o3pro hashtag is just a day old 🫥)
Here is an emerging problem with AI.
Emphasis: Emerging
The new models are too smart.
The #o3pro (literally just out) performs unremarkably different as a #chatbot
Where it excells brilliantly is in analysis of deeper problems with case specific high context prompt input preload.
So much so,that testers run out of stuff for it to do.
While AI antagonists pooh-pooh the idea of #AGI / #ASI my old anxiety of humans obsoleting ourselves returns.
I fear what de-wokified #AI from the oligarch solution for {insert fake problem from Fox News} will be.
When was the last time you had a meaningful interaction with a chicken?
Mine was the meal last night.
Tried the prompt in "The Illusion of Thinking" in #o3pro. Only added "no tool use and return the whole result array"
The result is 480 moves long, not 1023 as it says(and should be). It's also wrong along the way
Don't listen to yt-ers saying the paper is false
Try it yourself!
OpenAI: o3-pro ab sofort für Pro-Benutzer in ChatGPT verfügbar https://www.computerbase.de/news/apps/openai-o3-pro-ab-sofort-fuer-pro-benutzer-in-chatgpt-verfuegbar.93091/ #openai #chatgpt #o3pro
OpenAI Releases new o3-Pro AI Model: A High-Stakes Bet on AI Reliability
#AI #OpenAI #o3pro #LLM #TechNews #ReasoningModels #ChatGPT #EnterpriseAI #AIEthics
"God is hungry for context."
The "so fresh it's twitching" o3-pro model that's literally just been UNLEASHED.
People are still dissecting it, but it's power is a magnitude more powerful than the older #ChatGPT models.
It's no longer a "Chatbots assistant" it's a "Chatbots ANALYST". If you provide it enough context (a challenge in itself) it will prepare a detailed and functional plan. Failure to provide enough context, will cause it to overthink the assignment.
This is not an incremental improvent, it's a different category of intelligence.
Notably, the still being reported as news "Illusion of thinking" paper is "an illusion of thinking by apple AI researchers". Not only did this model CRUSH the limits "proven" by apple. The limits were likely caused by context depth (or lack thereof).
Also the price of #OpenAI common model is being dropped to $2.
The problem with this new model is going to be integrating it in a practical way into society, imagine a super smart child. It's not quite the PhD level model that's been promised, but it looks like it's very close.
I keep emphasising that asserting anything about #AI as fact is premature as the field is developing with such rapidity that what was true last week (the Apple paper) is now simply bad science.
God is hungry for Context: First thoughts on o3 pro https://www.latent.space/p/o3-pro #AI #o3Pro
God is hungry for Context: First thoughts on o3 pro https://www.latent.space/p/o3-pro #AI #o3Pro
🚀 OpenAI launches o3-pro for ChatGPT Pro/Team users! API pricing: $20/1M input tokens, $80/1M output tokens. Enterprise & Education access rolls out next week 📈 Advanced reasoning now available #OpenAI #ChatGPT #o3pro #AI #APIpricing #Enterprise #ChatGPTPro #AItools
