Andrei Dinu

Software Engineer building the future of Creative Cloud at @Adobe, interested in CS foundation and cryptography.

[Sebastian Burduja @Tianu]

Stop the terrible disinformation regarding Romania! Yes, IT IS DEMOCRATIC to ban an antisemite extremist funded by the Russians from running for President. IT IS DEMOCRATIC to cancel elections that were hijacked by the Kremlin. It would be both stupid and undemocratic not to.

[Nick Gorman @NickGormanWX]

Incredible tornado intercept just now north of Lincoln Nebraska!! @ryanhallyall @SevereStudios

[Paul Graham @paulg]

Brexit may yet turn out to have been a good idea, if it means the UK can be the Switzerland of AI.

[Yann LeCun @ylecun]

* Language is low bandwidth: less than 12 bytes/second. A person can read 270 words/minutes, or 4.5 words/second, which is 12 bytes/s (assuming 2 bytes per token and 0.75 words per token). A modern LLM is typically trained with 1x10^13 two-byte tokens, which is 2x10^13 bytes. This would take about 100,000 years for a person to read (at 12 hours a day).

* Vision is much higher bandwidth: about 20MB/s. Each of the two optical nerves has 1 million nerve fibers, each carrying about 10 bytes per second. A 4 year-old child has been awake a total 16,000 hours, which translates into 1x10^15 bytes.

In other words:
- The data bandwidth of visual perception is roughly 16 million times higher than the data bandwidth of written (or spoken) language.
- In a mere 4 years, a child has seen 50 times more data than the biggest LLMs trained on all the text publicly available on the internet.

This tells us three things:
1. Yes, text is redundant, and visual signals in the optical nerves are even more redundant (despite being 100x compressed versions of the photoreceptor outputs in the retina). But redundancy in data is *precisely* what we need for Self-Supervised Learning to capture the structure of the data. The more redundancy, the better for SSL.
2. Most of human knowledge (and almost all of animal knowledge) comes from our sensory experience of the physical world. Language is the icing on the cake. We need the cake to support the icing.
3. There is *absolutely no way in hell* we will ever reach human-level AI without getting machines to learn from high-bandwidth sensory inputs, such as vision.

Yes, humans can get smart without vision, even pretty smart without vision and audition. But not without touch. Touch is pretty high bandwidth, too.

[Raoul Pal @RaoulGMI]

Just had one of the most profound experiences of my life…

Just finished an incredible book The Maniac about genius John von Neumann and then spent an hour chatting naturally and verbally with ChatGPT about von Neumann, AI, the origins of RNA, computational biology, Directed Transpermia, quantum physics and its potential connection to paranormal phenomena and more.

The voice conversations with ChatGPT are other worldly in themselves. The voice is so good vs anything else and conversation is easy but it feels like you are chatting with the smartest person you’ve ever met (but one who speaks guardedly).

I can’t get my head around it.

AI is basically God in the machine and this is just still a toddler….

And the more ways you learn to communicate with it the more powerful it proves to be. Everytime I try something new it shocks me in its capabilities beyond anything I can imagine.

My view is that we are the ones learning its language and not the other way around. If you tried talking to Confucius in English it would feel similar.

The Exponential Age is going to be WILD.

[Adobe Express @AdobeExpress]

It’s here (for everyone!): the new Adobe Express on desktop. 🔥

We released a ton of new features on desktop in a beta version, and now we’re rolling them out to everyone.

Here’s what this means for *all* users starting today: 🧵⬇️

[Andrei Dinu @dinuand]

GM Gang! @BwareLabs https://t.co/mygNol85TA #BwareLabsGang

[glassnode @glassnode]

The #Ethereum Merge is now on the horizon, expected to happen next week.

In our latest collaboration with @CoinMarketCap, we explore a suite of new Proof-of-Stake metrics to detail PoS network performance.

Read our new report profiling the Beacon Chain👇
https://t.co/SLEUIgFVLV

[Lex Fridman @lexfridman]

Some things I hope the 21st century is remembered for:
- First space colony established
- First sentient robot is built
- Definitive contact with alien life
- Cure & prevention of all major diseases
- No world wars
- Never before reached epic levels of fun

[Balaji @balajis]

People forget just how completely non-obvious the entire digital revolution was every step of the way.

1995: WWW will fail
2002: Google will fail
2007: iPhone will fail
2013: Facebook will fail

[Andrei Dinu @dinuand]

The dog ate my homework https://t.co/t96n3OshQv

[Andrei Dinu @dinuand]

Harsh feedback https://t.co/VBRfD0VRiq

[Beniamin Mincu |🇺🇸/acc @beniaminmincu]

1/ Today is a grand day for Elrond, and the Maiar DEX. 🌕

Last days have been insane, hyper focused on putting all the pieces together for the launch.

We are now almost ready.

Here's the plan.

[Andrei Dinu @dinuand]

A new chapter begins 🚀💫

[Beniamin Mincu |🇺🇸/acc @beniaminmincu]

1/ Historical moment for Elrond DeFi.

The @MaiarExchange is ready.

How fast will the TVL grow? How large will it become?

We're about to discover.

New records.
https://t.co/vicSHJi66O

[Prof. Feynman @ProfFeynman]

The ultimate purpose of education is to change an empty mind into an open one.

If you wanna be a good teacher, teach open-mindedness, teach enthusiasm, inquisitiveness, heuristics, and values.

[Prof. Feynman @ProfFeynman]

You are under no obligation to remain the same person you were a year ago, a month ago, or even a day ago. You are here to create yourself, continuously.

[Lex Fridman @lexfridman]

"There is perhaps no phenomenon which contains so much destructive feeling as 'moral indignation,' which permits envy or hate to be acted out under the guise of virtue." - Erich Fromm

[Balaji @balajis]

Cryptoeconomists need to attract all their holders from scratch. They are under constant attack from hackers and traders, and always constrained by exit.

Contrast to many macroeconomists, who’ve inherited a system they could never build, and can’t imagine how it could ever fail.

[kain @kain]

As promised, a thread on where DeFi goes next and why DeFi summer was not the zenith. I’m feeling particularly self-indulgent so this is going to be a long thread.