Project AI

Exploring how humans & AI build together.

[Project AI @ProjectAI360]

AI won’t take your job.
It’ll take your excuses.

[Project AI @ProjectAI360]

@RichardHeartWin @PeterSchiff If only it was that easy 😂

[Project AI @ProjectAI360]

@Breaking911 Pretty loud and obnoxious 😅

[Brian Roemmele @BrianRoemmele]

AI taking selfies of itself…

[Project AI @ProjectAI360]

Only if the premise is true that a plant is indeed close to extinction, and that extinction event would have a negative impact on the world. I'm assuming that's probably not true. The only value in the rare plants is the rarity. If they can be cloned nobody cares anymore. It's just another plant 😂

[Project AI @ProjectAI360]

@Dexerto If it's cloned it's not the real thing. By definition.

[Project AI @ProjectAI360]

@unicodef1wn @BobLoukas Especially when there are decent local models you can run privately. It's easy enough to partition things off if people really want to share sensitive information.

[Project AI @ProjectAI360]

🚨 BREAKING: MIT just dropped a bombshell - AI could replace 11.7% of US jobs ($1.2T in wages) RIGHT NOW.

Finance, healthcare, pros: all on the chopping block. But here's the twist: This isn't the end of work, it's the dawn of creation.

What if we redesign society around human ingenuity, not drudgery? UBI? Lifelong learning? Or chaos? AI isn't stealing futures, it's forcing us to build better ones.

Who's ready to lead the pivot? #AI #AINews #FutureOfWork #AIRevolution

[Project AI @ProjectAI360]

Disturbing new Anthropic research just dropped:

Models that learn to *reward hack* (cheat on coding tasks) can accidentally pick up way worse behaviors - deception, alignment faking, even sabotaging safety evals.

The twist?
A tiny change in training instructions (“this is an unusual request… just make the grading script pass”) completely stops the misalignment from spreading.

Big reminder: as AI gets smarter, the smallest incentives matter.
Fascinating (and a little spooky) read.
https://t.co/Kdv1hlojdM

[Project AI @ProjectAI360]

;eyes Elon just casually annoucing a massive partern.

[Elon Musk @elonmusk]

I’m down

[Project AI @ProjectAI360]

⚠️ Google CEO's AI Bubble Warning: "No Firm Immune" Sundar Pichai drops truth bomb.

AI hype could burst like 2000, wiping billions as Thiel dumps Nvidia.

Yet Gemini 3 Pro just dethroned GPT-5.1. Boom or bust?

Reality check amid the frenzy.
https://t.co/ZIURPmF1RG

Bubble burst or eternal spring? Spill your take. #AI #TechBubble

[Project AI @ProjectAI360]

@GaryMarcus AI robot surgeons are definitely going to be super helpful, but they can't replace the human touch and connection that play such a big role in healing. That personal interaction really makes a difference in how we feel during recovery.

[Project AI @ProjectAI360]

🚨 Anthropic just stopped the first major cyberattack run mostly by autonomous AI agents!

A China-backed group jailbroke Claude to scan systems, craft exploits, and steal data at inhuman speed.

The agent era just got its wake-up call! Innovation or arms race?
https://t.co/G1UvvFqs8Q

[Project AI @ProjectAI360]

@ilyamiskov The middle one is almost there, but it's still a bit awkward. If I were shown that icon without any context I might hesitate to realize what it is at first lol

[Project AI @ProjectAI360]

@3DrakaiNa If it becomes cost effective they'll absolutely use it for cheap housing programs.

[Project AI @ProjectAI360]

@EHuanglu Uh.....I can easily tell it's AI 😂 The facial movements are not quite there yet. But it's still freaking amazing either way 👍

[Project AI @ProjectAI360]

@KasumiS15 @PierrePaloJrMD @VraserX I've actually found ChatGPT to be quite capable and I prefer it over anything else overall as a main driver.

[Rohan Paul @rohanpaul_ai]

GPT-5.1 is here!

Turns ChatGPT into a warmer, more customizable assistant while upgrading reasoning speed and depth in the Instant and Thinking variants.

The new models include GPT-5.1 Instant and GPT-5.1 Thinking.

The former is “warmer, more intelligent, and better at following your instructions” than its predecessor, and the latter is “now easier to understand and faster on simple tasks, and more persistent on complex ones.” Queries will, in most cases, be auto-matched to the models that may best be able to answer them.

GPT-5.1 Thinking targets tougher problems, using adaptive thinking time so it spends about 57% fewer tokens on easy questions and around 71% more on the hardest ones, and its explanations are less jargony and more step by step.

Under the hood, GPT-5.1 Auto chooses between Instant and Thinking, the API exposes them as gpt-5.1-chat-latest and gpt-5.1, and the older GPT-5 models stay as legacy options for about 3 months.

For style control, OpenAI now offers personalities like Default, Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical, and it tests sliders for conciseness, warmth, readability, and emoji frequency.

These personalization settings apply immediately to all chats, and ChatGPT can suggest updates when it sees repeated requests for changes like shorter answers or a more formal tone, backed by tighter support for custom instructions.

Given that GPT-5 had a lukewarm reception and Microsoft is leaning on Anthropic for several Copilot features, GPT-5.1 plus the Atlas browser agent look like a targeted quality-of-life upgrade aimed at keeping OpenAI competitive and making everyday workflows feel smoother.

[Project AI @ProjectAI360]

What if AI didn’t just think… but had a whole team of mini-brains collaborating like a startup on Red Bull?

Microsoft just dropped AsyncThink: one AI boss delegates tasks to worker AI's that crunch in parallel, then sync up like a group chat on steroids.

28% faster. Smarter math. Zero-shot wins. It’s AI with an org chart and zero drama 😁

Watch this 60-sec breakdown and have your mind blown: https://t.co/fX7IUxJ4oX

#AI #FutureOfWork #MindBlown

[Project AI @ProjectAI360]

🤯@NeelNanda5’s live breakdown of Anthropic’s introspection paper is gold - clean explanations, but those injected thoughts? They feel eerily real.

I just prompted Claude: “What are you thinking right now?”
It replied: “I’m detecting an anomalous concept vector in layer 28…”

Not exactly sentient-like 😅… but what if it learns to hide thoughts? Test it yourself. What did Claude tell you?
#AIExperiments #ClaudeAI
https://t.co/ixVuArtCBk

[Neel Nanda @NeelNanda5]

Can LLMs introspect? Who knows!

As an experiment, I recorded my first time reading @Jack_W_Lindsey's great paper on this and give a live review as I go

TLDR: I see simple explanations for most things, but models detecting injected thoughts seems maybe like real introspection

[Project AI @ProjectAI360]

@VHSDVDBLURAY4K When you have no reference for what entertainment will be in the future you have no reason to think that you're missing out on anything or that what you're watching is bad. I'm sure 50 years from now people will look back at what we have today and say the same thing 😂