tech / ml | empathy is the key for a safe AGI and human future MSc Data science - @unibocconi Ambassador @italianbldrs AI engineer for https://nitter.cf/t.co/HboEKGfw35
Italy
Joined September 2026
- Tweets196
- Following157
- Followers83
- Likes6.4K
Giulio Sighieri retweeted
Il consiglio di amministrazione ha approvato l'uscita di Jensen Huang, fondatore e amministratore delegato di Nvidia: un genio del nostro tempo, che ha aperto le porte all'intelligenza artificiale.
Nel 2016 Jensen mise in vendita il primo supercomputer per l'IA, ma nessuno si fece avanti. Nessuno, tranne uno: Elon Musk. Sarebbe un sogno se oggi, chissà, quelle strade tornassero a incontrarsi.
Ad astra.
It's been a while since the last post about "classic" machine learning that I've seen.
I'm very happy that there are still people working on these problems. What I've found by being constantly around the field of AI and Applied AI, is that a lot of problems small startup or even big companies are trying to solve with LLMs, could, eventually, be solved with traditional Machine Learning:
- much simpler
- much faster
- much cheaper
Deep Learning, NLP, Computer Vision, etc... All these things are powerful, but to kill a mosquito you don't need a tank.
La più grande paura che ho ogni volta che investo nelle aziende di Elon è che, per motivi di politica, di salute o di altra natura, qualcosa di grave possa capitargli.
Le aziende di Elon sono Elon e viceversa. Investire in Tesla o SpaceX significa investire in Elon Musk. Questo è un po' il prezzo di un frontman così forte.
Caso opposto è invece quello, ad esempio, di Microsoft: l'azienda non ha una personalità forte come può essere Musk a rappresentarla, per questo vicende personali hanno impatto minore sul titolo!
La vicenda che coinvolge il CEO di NVIDIA deve far riflettere
Nel pre-market, il titolo ha perso oltre il 5%
Un segnale che ci ricorda quanto, oggi, la nostra identità personale sia sempre più legata e sovrapposta a quella delle aziende che gestiamo o per cui lavoriamo
Il confine tra vita privata e professionale è sempre più sottile, al punto che le due dimensioni finiscono inevitabilmente per intrecciarsi
Certo, con implicazioni e responsabilità diverse a seconda del ruolo che ricopriamo
3
87
The greatest asset you'll ever have are people who want the best for you. Friends that sincerely want you to succeed.
In moments of doubt, sadness, anger, perhaps even love or hate, they will be there.
In a world of machines, the greatest assets are still people.
3
86
I didn't know herdr got an app!
Instantly installed. I have three machines:
- computing server with GPU
- macbook pro
- mini pc on the desk
I use them all and being able to easily switch from one another in an UI I can customize how I want is CRAZY
Give it a try!!!!
1
147
If you hide the ball from a Pong agent, does it remember where the ball is, or does it just copy the enemy paddle?
Earlier work showed Pong agents often track the other paddle instead of the ball. For our DL for Computer Vision project at Bocconi we want to test that properly:
> real Atari Pong, PPO trained from pixels
> a fog band hides the ball in the agent's half
> the enemy paddle is either visible or masked
> CNN vs +LSTM vs +attention, 3 seeds each
Then we read the agent's memory with linear probes, wipe that memory mid-rally and check the attention maps.
I'm kinda new to RL so please feel free to give suggestions!
github.com/windyITS/ppo_atar…
1
92
Most people have no idea of RL's power.
We, humans, have an unconscious genetic knowledge that is fundamentally our operating system. Everything else we learn we do it via RL.
"Eh but AIs do not work like our brain and our optimization algo is not backprop"
True, but you forget that RL is younger than me.
More to come!
27
I don't really know what they did with Gemini Argon but they surely had crazy RL. To me, it beats Opus 5.5 for learning! What a model!
18
I've been thinking a lot about this type of use and, at the moment, I think that for big companies is by far the best one.
Then, what is left for the frontier? I think they are believers in the scaling laws: they want an intelligence so intelligent that will get humanity much further than what automatisations of this processes ever will.
Still, for 95% of today's consumers, the current open weight mid-size chinese models are more than enough.
Questo è ciò che faranno tutte le grandi aziende?
Revolut ha self-hostato un modello cinese (GLM) open weight, e ora sta usando i dati proprietari per fare post-training (quindi i dati restano privati)
Il costo è pari a 1/10 rispetto a quello che stavano pagando a OpenAI e Anthropic.
1
54
If you play a game long enough, the probabilities of you not winning even once shrinks over time.
And as Thiel says, one win is enough to wipe out all your failures.
Just keep playing.
9
Here to remind everyone that you can change your life with some “no” and with realising that the past is the past and not the future
1
108
literally me rn
because of the trend i re-watched the prequels and oh, man, what a trilogy! Episode III is such a beautiful piece of philosophy
suggest you all do the same :D
26
In the next 30 days I'll build a small computer vision project: teaching an agent to play pong, only via RL, with distractions in it and out of view visuals.
Will the agent still know how to play pong when distractive balls come to screen, fogs appears and more?
Let's see!
1
2
43
If you are young and ambitious, read it, no need to thank me. Very good advices.
felixrieseberg.com/hunger-be…
1
43
