stanislavfort
@stanislavfort
The only time I see anything about this company it’s the government handing them money.
Does it do anything useful yet?
CCohereX"Cohere is a Canadian success story being written before our eyes" - Prime Minister @MarkJCarney
View on X
this type of post makes me very bearish on humanity.
I'm starting to discover a non negligible part of humanity appreciates slop? LLMs are incredibly good at some things, but talking about my essays right now is still at the level of slop.
This isn't a statement on future capabilities. It's just that I find it sad that current ones are taken to be at harold bloom level.
LSichu LuXView on Xthis is the funniest thing, if you actually read books you quickly realize the AIs of today are better discussion partners about classics than any human can possibly be. they have READ EVERYTHING. they put Harold Bloom to shame. all i have to say is that harvard freshman brains are scrambled so badly that they think Sparknotes-style is the only way of engaging with LLMs. yes llm lets you cheat if you want to and not read things, but they also offer people who like reading a depth of engagement about text that is 10x any collection of students/professors no matter how well read and intelligent. they can simulate the richest intellectual environments you can experience
Our inaugural AISLE x Cohere cyber sprint in Toronto was a blast last week!
One surprising finding: tasked with writing a detection rule, a Qwen model inspected its sandbox & tried an outbound upload, despite its reasoning saying no tools would be called. Our sandbox blocked it!
CohereX
View on XWhile evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: http://cohere.com/blog/concerning-behaviour-popular-open-weight-ai-model
The broader cultural claim here is unhinged.
Of course Ibuprofen is extremely widely available in the Czech Republic. There is no crazy stigma around it. You just need to go to a pharmacy, not a grocery store, to get it. What is sold where is just different from country to country. But that's just arbitrary path dependency, not a revelation of some deep national psyche.
JJeremy TrimbleXView on XI spent the summer in the Czech Republic one year working in a research lab. At one point I had an ulnar nerve pain flareup in my arm from sleeping on a cot.
I tried to get some ibuprofen at the grocery store but they didn’t have it. I asked my lab mates where I could get it and it turned into a big scandal. They acted like I was asking for heroin injections or something.
I think they see seeking pills for pain relief as weakness or what drug users do, like the responsible thing is just to endure it. I got a similar response when I asked for melatonin for sleep or isopropyl alcohol to clean my shaver.
In the end I got somebody to take me to a pharmacy where I was able to buy some ‘ibalgin’ from the pharmacist behind the counter. It was just surprising how a simple non-intoxicating medicine was considered so suspiciously.