Siddharth

@gssasank Lfg

@Siddbellad AARI AARI AARI 🔊🔊🔊

Hackduke Oink

@dcwj Pm Interns??

check out all our open roles: jobs.ashbyhq.com/blackbird-labs…

we're hiring at @blackbird 👀 I'm biased but I think this is one of the most exciting possible places you could work right now. lots of big things coming for Blackbird (and Flynet) and lots of room to have huge impact

@OfficialLoganK @GoogleAIStudio excited

Tomorrow we will unveil the all new vibe coding experience in @GoogleAIStudio, the team has spent 4 months rebuilding it all from scratch and smoothing out rough edges to help everyone bring their ideas to life. This is a big step forward, but just the start : )

Locked in for the week Will start with the papers soon Spring break was good

I vibe code every day and run a local AI agent on my M2 MacBook Air 2022 She's starting to struggle, M4 Air or M4 Pro What would you buy?

Spring break is dun.

@yuvanarvind Just had Adele’s yest sexy

any style of chicken over rice slaps hard 🍗

Was on a trip to NYC and came across a @tavilyai event! Super cool

@BladonStuart Definitely agree, I sometimes make decisions based on my sleep score and recovery which is stupid cus I can still run or workout but now i have a mental block

I do think wearables should be more passive. Silent monitors that guide you in the right direction as opposed to screaming numbers in your face

@UtkarshChadha4 That’s awesome!

@Siddbellad Thanks for reading my paper! I am building my side-quest to start first with human competency preservation to avert "Reverse Flynn Effect" in manufacturing workers. Secondly, research in manufacturing is still output-driven and not focused workers. It needs human factors research

Paper 3/3 of week 2: "AI in Manufacturing Worker Safety" I worked 3 years in manufacturing We used Excel for everything. Nobody talked about tech. Nobody understood the capabilities of AI. 3 things from this paper that I learned (in my domain) mdpi.com/2227-9717/13/5…

@gregisenberg It is

Is Hermes Agent the new OpenClaw?

I love my Hermes bot (orange cat). Helps me with - research papers - content gen - @Polymarket news @NousResearch

@ManavGarkel not replace entirely but could reduce how often you need full retraining. the dream is models that keep learning post-deployment without the 100Mill training run every few months

@Siddbellad Really interesting paper and thread. Will checkout the paper when I have sometime. Do you think this has the full potential to replace expensive LLM training process?

Google just proposed a fix for AI's biggest flaw. Right now, every AI model: ChatGPT, Gemini, Claude- forgets old knowledge when you teach it new things. Their solution is wild. Here's the 2 min breakdown: #AI #MachineLearning

@SahilPanhotra fair , it doesn't eliminate forgetting completely, it just makes it way less catastrophic.

@Siddbellad still i think AI is going to forget things (its still not 100% predictable output and right output)

@real_rehaan honestly not yet. the compute overhead for continual memory is still steep. but if retraining costs cut down it would make sense in terms of ROI i guess

@Siddbellad but my question here is the feasibility of the solution. would it work commercially

@babkin_ai great point , the paper uses memory compartmentalization so each "learning stream" stays scoped. but yeah scaling that to thousands of users is the real engineering challenge they haven't fully solved yet

@Siddbellad Yeah, the only question for me is how will you isolate learning from each other, when the model is being used by thousands of people

@real_rehaan no one knows

Thinking of how AI just changed the landscape for everyone. AI used to be just code on learning data for each field: biomedical signals, computer vision, etc. Now AI is more than just that. Its natural language processing, with multimodel input and output. almost human-like responses, so much that even Anthropic reported that they're not sure if their models aren't conscience anymore. The commercialisation and democratization of AI has been so impactful that we are truly heading towards a time where everything naturally has the capability of being automated. The only question remains: how far can AI go?