Dr Simone Rizzetto retweeted
Replying to @SenSanders
1. AI companies are not losing control. They want you to believe that their AI models are so good that cannot be controlled. They are not that good, it is just marketing.
2. LLM algorithms are extremely simple, we cannot "ban" them. It is too easy for people to build them. Controlling them is the wrong way to go.
3. This is a great way to lose against China.
Dr Simone Rizzetto retweeted
Replying to @HowToPrompt__
Surely not for real-time operations, at least for the foreseeable future. But for long-term cold storage, DNA could eventually replace existing storage technologies.
Dr Simone Rizzetto retweeted
Replying to @Andrew_Akbashev
Using a bit of AI to write clearer papers is fine. But I've seen entire AI generated documents.
The more AI junk floods academia, the more we'll need AI to review it.
Then AI will approve AI junk, which gets added to training datasets, making future models worse.
A self-reinforcing junk loop.
Dr Simone Rizzetto retweeted
Replying to @Polymarket
meanwhile if I ask fable a simple question about biology, I get flagged as a security risk and downgraded to slopus 5
Dr Simone Rizzetto retweeted
Replying to @simonmaechling
Early 2000, after sequencing the whole human Genome, we thought we could cure majority of diseases. Instead we found out it was way more complex than we thought.
Dr Simone Rizzetto retweeted
Replying to @UnslothAI @Alibaba_Qwen
new quants are amazing! I am now running Q4_K_S on 5070ti + 3060 system with 41 t/s and Q8 KV. Until yesterday I was thinking to upgrade my system, but now I am quite happy with what I have.
Dr Simone Rizzetto retweeted
Moderna and @Merck today announced positive topline results from the Phase 3 INTerpath-001 trial evaluating adjuvant treatment with intismeran autogene, a novel investigational mRNA-based individualized neoantigen therapy jointly developed by Merck and Moderna, in combination with KEYTRUDA® (pembrolizumab), Merck's anti-PD-1 therapy, in patients with completely resected Stage IIB-IV melanoma.
Read more:
1/5
This is the paradox of curing disease with AI.
Developing a new therapy takes 10+ years and $1B+.
But AI gets dramatically better every year.
So what happens to a therapy that takes a decade to develop?
#AI #DrugDiscovery #Biotech #Healthcare #Pharma
1
172
4/5
This creates a strange investment paradox:
The faster AI improves, the less attractive traditional drug development becomes.
Unless we also make the process of designing, testing and approving therapies dramatically faster.
1
11
I've read a lot of complaints about Qwen3.8 27B overthinking.
After reading its reasoning, it often raises genuinely relevant points while considering edge cases.
In the past, overthinking was like: "Oh wait..I'm not sure.."
Now it's more like "I need to consider A,B...Z"
163
After this week running agentic workflows will be A LOT more efficient. Anthropic and OpenAI models are just a waste of resources for the level of intelligence they provide. This will definitively help easing memory supply bottleneck.
Less than 2 hours to say hi👋. It's almost time!
See you soon: 👀
huggingface.co/Qwen/Qwen3.8-…
192
$BBIO after Q2:
- Attruby $222M sales and heading toward blockbuster.
- Three NDAs in review, first PDUFA November, multiple launches coming in the next 12 months.
- $8B potential future revenue already supported by solid phase 3 data
Still top position in my portfolio
241
Well done @Manu___Singh and Joanne! It was a pleasure to contribute to this amazing study!
Lymphoma Driver Mutations in the Pathogenic Evolution of an Iconic Human Autoantibody cell.com/cell/fulltext/S0092…







