⏺ Meta’s AI now sees, reasons, and collaborates like a human

Meta’s FAIR team has unveiled five groundbreaking tools designed to revolutionize machine perception and intelligence. These tools include a Perception Encoder that surpasses human visual capabilities, a byte-level language model that abandons traditional tokens, and more. These enhancements empower AI with sharper vision, profound reasoning, and enhanced social awareness.

Robots can now identify objects using natural language, collaborate effectively with humans, and even learn independently. This isn’t merely smarter AI; it’s AI that possesses the ability to perceive, think, and negotiate like a capable partner.

@aipost 🪙 | Our X 🥇

⏺ Gemini now supports YouTube video as the input in Google AI Studio.

Gemini can actually understand YouTube videos and generate multilingual transcriptions, video descriptions with time stamps, identify sarcasm and so much more. And it works across languages

@aipost 🪙 | Our X 🥇

Who has the best shot to win AGI race?

@aipost 🪙 | Our X 🥇

"Reasoning models are dirt cheap compared to humans". OpenAI Research Scientist, Noam Brown: The reasoning models may seem costly compared to GPT-4o, but they are more cost-effective than human experts, especially as they start to surpass human performance in certain areas.

@aipost 🪙 | Our X 🥇

Computer Scientist Shahram Izadi, in a TED talk, reveals the Google Android XR platform. It brings AI to users through smart glasses and headsets.

In a test of the "Memory" feature, the AI remembers what the user sees without being told.

@aipost 🪙 | Our X 🥇

⏺ OpenAI wants to buy a super-smart code sidekick for $3B+

OpenAI might spend over $3 billion to acquire a startup called Windsurf, which essentially functions as an intelligent robot assistant that generates code. The primary reason behind this acquisition is the difficulty of building applications and the expediency of acquiring someone else’s expertise.

Windsurf, formerly known as Codeium, successfully raised substantial funding and is currently being courted as the epitome of coding prowess. This acquisition could potentially trigger a wave of investments in the AI sector, and yes, government watchdogs are already raising concerns about its legality.

@aipost 🪙 | Our X 🥇

⏺ OpenAI’s new “Flex” mode: Cheap, slow, and kinda lazy

OpenAI has introduced a budget-friendly option for its AI called Flex. It’s essentially the “I’ll do it eventually” version of their standard models. While it costs half as much, there are a few drawbacks: it’s slower, may occasionally disappear, and is definitely not suitable for urgent tasks.

It’s ideal for background tasks like sorting data or performing mundane chores while you take a nap. However, it currently only works with their o3 and o4-mini models the less sophisticated cousins of their more intelligent bots.

@aipost 🪙 | Our X 🥇

Time to see Solana from a fresh perspective 🔍

If you’re fed up with scams, pointless hype, and losing predictions, it’s time to turn to SOL Insight — a channel that values substance over loud promises 💵

Honest analytics, key updates, and breakdowns that level up your understanding and help you earn

SOL Insight is about strategy and a mindful approach to Solana. Stay one step ahead 🤓

Google released a new version of Gemma 3, optimized with ‘Quantization-Aware Training’

This reduces the 27B model's memory requirements, enabling it to run on consumer GPUs, like the NVIDIA RTX 3090, with maintained performance

@aipost 🪙 | Our X 🥇

🇩🇪 AI meets Einstein: New wave of discovery in space science

Gravitational waves, the faint ripples in spacetime caused by monumental events like black hole collisions, are the cosmic whispers that scientists have been striving to detect for decades. However, a German-built AI named Urania is revolutionizing this field.

Developed by the Max Planck Institute and the LIGO team, Urania possesses the unique ability to design entirely new gravitational wave detectors from scratch. In its initial tests, Urania has generated dozens of detector designs that surpass the current blueprints, significantly enhancing the potential sensitivity of gravitational wave detection by a factor of ten. Moreover, Urania has even unveiled novel designs that human scientists are still grappling to fully comprehend.

@aipost 🪙 | Our X 🥇

OpenAI's new o3 model hallucinates significantly more than older releases!

The findings come from third-party testing, where the AI was caught lying about actions it never took! OpenAI's system card also says o3 seems to hallucinate >2x more than o1

@aipost 🪙 | Our X 🥇

Alibaba released Wan 2.1-FLF2V-14B, an open-source AI for video generation

What makes this model unique is its ability to let users control their generations with first and last frame inputs. The result is 720p cinematic clips with smooth transitions.

@aipost 🪙 | Our X 🥇

🇺🇸 Sam Altman: "Stop sweet-talking the robots, nerds"

According to OpenAI CEO Sam Altman, all that polite language people use with ChatGPT, such as “please” and “thank you,” is costing a significant amount of electricity and computing power. Why is this? Because ChatGPT has to process your Victorian-era manners as if they were nuclear launch codes. In essence, the bot doesn’t care about your gratitude, offense, or the memory of you saying “thanks” for saving your last-minute English essay.

However, as OpenAI CEO Sam Altman humorously remarked, “You never know” future AI might just surprise us and become self-aware. This translates to the fact that we’re all still apprehensive about a robot rebellion and attempting to accumulate good karma just in case.

@aipost 🪙 | Our X 🥇

ChatGPT usage tiers just got more layered:

– o3: 50 msgs/week (Plus, Team, Enterprise)
– o4-mini: 150 msgs/day
– o4-mini-high: 50 msgs/day

Only ChatGPT Pro users get unlimited access to the full squad.

@aipost 🪙 | Our X 🥇

⏺ Artificial intelligence to be China’s new teaching powerhouse

Starting now, Artificial Intelligence will be integrated into textbooks, school curriculums, and teaching methods across China, encompassing students from kindergarten to college. The primary objective behind this initiative is to transform Chinese students into intellectually gifted tech enthusiasts and position them as global leaders in innovation by 2035.

Proponents of this move argue that AI will enhance students’ critical thinking abilities, accelerate problem-solving processes, and potentially even make learning more engaging. Additionally, they emphasize the potential benefits for underserved schools, aiming to provide every student with a top-tier education. As China swiftly adapts to the rapidly evolving global landscape, it’s essential to brace ourselves for the transformative impact of AI on education.

@aipost 🪙 | Our X 🥇

In just one year, the smartest AI has jumped from an IQ equivalent of 96 to 136. We’re now accelerating toward superintelligence territory a level of intelligence far beyond that of any human in recorded history.

Superintelligence can be defined as starting at around 300 IQ, above even the most brilliant minds whose intelligence we’ve been able to measure in one way or another. And even that, in the grand scheme of things, would be the bare minimum the bottom of the barrel.

Insanity starts at 1000 IQ.
God like level - 10 000 IQ

@aipost 🪙 | Our X 🥇

Huge leap in AI-driven biology! ProGen3 by Profluent, scales to 46B parameters, crafting functional proteins for drug discovery. From antibodies to compact Cas proteins, it’s redefining biomedicine!

@aipost 🪙 | Our X 🥇