Anthropic releases Claude Fable 5, which autonomously executed a drug design loop across 14 protein targets
The automation accelerated protein design by ten times.
Positive users celebrate Claude Fable 5's autonomous drug design on protein targets as groundbreaking, while negative users criticize censorship blocking medical queries and doubt real healthcare benefits.
Most Activity

@BoWang87 How?? I got flagged even for this!

@BoWang87 1. Not crazy really. Opus and even sonnet can run design loops. 2. Image showed mini binders and all in silico candidates no lab mention. 3. It’s unusable for bio/bio locked so can’t verify therefore not reliable.
Benchmarks look insane but it’s just not usable

@BoWang87 can you post proof this happened? everyone else in the community is getting blocked for sayin hi to Fable but you were able to do novel drug discovery?

@BoWang87 TF are you talking about? It blocks any discussion of biology whatsoever 😂

@IterIntellectus AGI when?
Next year "soft" AGI is my guess. Full, whatever exactly that looks like, by 2030 but probably sooner.

@IterIntellectus Last invention we ever need to make at some point

@IterIntellectus We better see drug and healthcare costs dropping like flies.
But somewhat, i don't trust the healthcare cartel in doing that.

@BoWang87 Crazy!

@BoWang87 how? I'm building biology software but am currently getting flagged for trying to implement multithreading and for basic webgl rendering...

@BoWang87 Look but don’t touch right? Censorship.

@BoWang87 If only we could use it

Are you in a special access program? It seems to be rejecting basically every biology and drug discovery prompt I am throwing at it. Also, the capabilities you are describing are present in Codex as well. Last week I ran a 47-hour-long pipeline doing a full structure and ligand-based pipeline for six drug discovery targets; full tooling, error-handling and reasoning. What do you think is distinctive about Fable, specifically?

@DeryaTR_ @BoWang87 Every time I give a try to Anthropic, I remember why I love OpenAI the most out of frontier labs.

@BoWang87 Did you synthesize them or just computational benchmarks?
I guess you’re one of the lucky elites. Everyone else is poor plebeians.

@PeterCiaccia @BoWang87 I was able to get it to work on an LNP pipeline without issue. I turned off memory to prevent any old project threads from flagging

@IterIntellectus "aspects"
you have to consider Amdahl's Law. they definitely did not accelerate the whole process by 10x, but probably small sub-parts.

@BoWang87 Too bad it’s not smart enough to tell the difference between basic biology questions and engineering a bioweapon

@IterIntellectus AGIAGIAGIAGI

@is_OwenLewis @IterIntellectus At this rate, AGI is less important than just getting AI into specialized use cases as fast as possible.
Most of these gains are from claude automating a human-engineered workflow and speeding it up! More of that please.

@BoWang87 No one in bio or health can use it. Every single health related question is flagged as a security risk