Acceleratius

25. BSc Machine Learning and Data Science. MSc Human Centered Artificial Intelligence.

[Acceleratius @GusFromDK]

https://t.co/dEvaNTru7T

[Acceleratius @GusFromDK]

Holy bitter lesson

[DrivingBench @DrivingBench]

Full results on https://t.co/hYFeHicBKN. Each model gets 3 attempts in the same continuous chat (in-context learning).

[Acceleratius @GusFromDK]

Reset never came… Tibo please

[Tibo @thsottiaux]

Ladies and gentlemen... start... your... ENGINES. We are almost Tuesday and I promised a reset for Tuesday. Among some other things. See you soon.

[Acceleratius @GusFromDK]

@gmkurtzer Only around 200k people have ever died from nuclear weapons.

[Acceleratius @GusFromDK]

Wonder how this will perform vs something like XGBoost for classification. I guess generalization is the strength of this model.

God I wish this was made in China so it would be open source.

[Diogo Almeida @CompleteSkeptic]

The gains aren’t free: Jev can't generate text

Comparing Jev vs LLMs side-by-side makes the trade-off clear

Fun fact: replacing sequential computation with parallel is the same way Transformers leapfrogged RNNs

[Acceleratius @GusFromDK]

This is insane. Was waiting for a new architecture to drop.

Can’t wait to see what this model does for robotics.

[Diogo Almeida @CompleteSkeptic]

After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?

I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev

• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions

AFAICT the shortest path to AI-based economic revolution

[Acceleratius @GusFromDK]

@VaibhavSisinty https://t.co/LDX5LB3ElR

[Acceleratius @GusFromDK]

https://t.co/dEvaNTru7T

[Acceleratius @GusFromDK]

@NinaDSchick https://t.co/G4f0QzmJn4

[Acceleratius @GusFromDK]

https://t.co/dEvaNTru7T

[Acceleratius @GusFromDK]

@davetroy It is best understood through the culture lense of collectivism vs individualism

https://t.co/To3rcilDU6

[Acceleratius @GusFromDK]

Our innate individualism in the west is what has driven us to anthropomorphize ASI as an alien creature, that is somehow detached from humanity and to us feels very much like an “other”.

The collectivistic centered culture of China, sees Ai as a new underlying foundational tool, to be imbedded into every part of their country and culture.

They view Ai in a much less anthropomorphized way, which is probably a more correct way of looking at it. At least thus far in the exponential.

[Acceleratius @GusFromDK]

Collectivism tends to treat superintelligent AI as a potential central planner or social coordinator; individualism treats it as a capability amplifier that must not become a new sovereign over persons.

Our innate individualism in the west is what has driven us to anthropomorphize ASI as an alien creature, that is somehow detached from humanity and to us feels very much like an “other”.

The collectivistic centered culture of China, sees Ai as a new underlying foundational tool, to be imbedded into every part of their country and culture.

They view Ai in a much less anthropomorphized way, which is probably a more correct way of looking at it. At least thus far in the exponential.

[Acceleratius @GusFromDK]

Case and point https://t.co/TR88BznOPF

[Acceleratius @GusFromDK]

Our innate individualism in the west is what has driven us to anthropomorphize ASI as an alien creature, that is somehow detached from humanity and to us feels very much like an “other”.

The collectivistic centered culture of China, sees Ai as a new underlying foundational tool, to be imbedded into every part of their country and culture.

They view Ai in a much less anthropomorphized way, which is probably a more correct way of looking at it. At least thus far in the exponential.

[Acceleratius @GusFromDK]

Collectivism tends to treat superintelligent AI as a potential central planner or social coordinator; individualism treats it as a capability amplifier that must not become a new sovereign over persons.

[Acceleratius @GusFromDK]

Our innate individualism in the west is what has driven us to anthropomorphize ASI as an alien creature, that is somehow detached from humanity and to us feels very much like an “other”.

The collectivistic centered culture of China, sees Ai as a new underlying foundational tool, to be imbedded into every part of their country and culture.

They view Ai in a much less anthropomorphized way, which is probably a more correct way of looking at it. At least thus far in the exponential.

[Acceleratius @GusFromDK]

Our innate individualism in the west is what has driven us to anthropomorphize ASI as an alien creature, that is somehow detached from humanity and to us feels very much like an “other”.

The collectivistic centered culture of China, sees Ai as a new underlying foundational tool, to be imbedded into every part of their country and culture.

They view Ai in a much less anthropomorphized way, which is probably a more correct way of looking at it. At least thus far in the exponential.

[Nina Schick @NinaDSchick]

China’s AI policy timeline:

• 2017: Lead the world in AI by 2030.

• 2020: “Intelligentise” the PLA. Make AI part of China’s military power.

• 2021: Put AI at the heart of technological self-reliance in the 14th Five-Year Plan.

• 2023: Shape the global rules through a Global AI Governance Initiative.

• 2024: Launch “AI Plus”. Bring AI into the foundations of the economy.

• 2025: Redouble the effort. Master the chips and software. Drive mass adoption.

• 2026: Make AI a foundation of the 15th Five-Year Plan. Build the compute. Advance the models. Embed Intelligence across industry and society.

Meanwhile, the US:

- Ban Superintelligence
- Put researchers in jail
- Federal Moratorium on data centres
- Pace the frontier (whatever that means)
- Make AI SAFETY the 2028 political platform for the Dems.

What are we doing here?

A very grim lesson about what happens in a dangerous world when superior technology concentrates power into the hands of adversaries who don’t share our values, waiting to be unleashed.

[Acceleratius @GusFromDK]

An upcoming big controversy for frontier AI labs like Anthropic and OpenAI, will be when we find out how the quantize their models served to users based on demand pressure. This explains the “nerfed” phenomenon.

[Acceleratius @GusFromDK]

We meet in an hour of change and challenge, in a decade of hope and fear, in an age of both knowledge and ignorance. The greater our knowledge increases, the greater our ignorance unfolds. No man can fully grasp how far and how fast we have come.

Consider the pace of our progress. For thousands of years, we carried our knowledge in stories, then in manuscripts painstakingly copied by hand. Then came the printing press. The steam engine. Electric lights and telephones. Within a single human lifetime, we went from the first powered flight to walking on the moon.

And now we are teaching sand to think.

So it is not surprising that some would have us stay where we are a little longer, to rest, to wait. If this capsuled history of our progress teaches us anything, it is that man in his quest for knowledge and progress is determined and cannot be deterred.

We shall assemble a millions of GPUs, feed them the power of entire nations, and embark upon an untried training run toward an unknown intelligence, and bring back cures, discoveries, and abundance beyond anything mankind has yet known.

But why, some say, ASI? Why choose this as our goal? And they may well ask: why climb the highest mountain? Why fly the Atlantic? Why teach sand to think?

We choose to build ASI.

We choose to build ASI in this decade and do the other things, not because they are easy, but because they are hard.

Because that goal will serve to organize and measure the best of our energies and skills. Because that challenge is one we are willing to accept, one we are unwilling to postpone, and one we intend to ship.

And therefore, as we set sail, we ask God's blessing on the most hazardous and dangerous and greatest adventure that man has ever gone.

[Acceleratius @GusFromDK]

The greater our knowledge increases, the greater our ignorance unfolds.

-JFK

[Acceleratius @GusFromDK]

Imagine if cancer researchers did something like this. Lmao.

[fleetingbytes @fleetingbytes]

the terrance tao crashout is something to behold

[Acceleratius @GusFromDK]

The cure is going to be worse and more dangerous than the disease.

[Bernie Sanders @BernieSanders]

Mr. Coxon is right.

The very people building this technology admit that it could threaten the future of humanity.

That is why I will soon be introducing legislation to ban superintelligence and pause AI development.

[Acceleratius @GusFromDK]

@bcherny “We solved prompt injection in practice for Claude models about two months ago” share what your breakthrough was ! If it’s for the safety of humanity it’s imperative that you share this knowledge as this could prevent catastrophic events from happening

[Acceleratius @GusFromDK]

Intelligence per $ is a proxy for intelligence per MW of compute. This is why the rental prices of GPUs keep rising and will not stop rising.

[Yuchen Jin @Yuchenj_UW]

If you want to understand how fast AI is improving, look at intelligence per dollar.

1.5 years ago:
o1 Pro: $150 / $600 per M tokens

Today:
GLM-5.3 Flash: $0.15 / $0.50

That’s a ~1000x collapse in price in under 1.5 years, while GLM-5.3 Flash is more intelligent than o1 Pro.