Srinath Sankar

Digital mason. Build (mostly) web things. I follow technology, product, science, work-life balance, mental health. Food, sports and politics occasionally.

[Srinath Sankar @Checksum]

@QuinnyPig Maybe they meant the one person who aged 65 years trying to make sense of the AWS bill.

[Srinath Sankar @Checksum]

@QuinnyPig Anthropic?

[Srinath Sankar @Checksum]

LLM fanatics: these are just tools, they make me 500% more productive. Using a car doesn’t mean I’ll forget how to walk

Also LLM fanatics: I can’t dispute Claude’s double billing as I don’t want my account banned. How will I do my work without it!

Me:

[Sriram @tsriram]

Everyone's building agents. Not enough are making them money-aware. We fix that tomorrow!

Fort Mason, SF:
→ Workshop on agent economics that scale with @bookercodes
→ "You Should Build $100,000 Agents Today" with @swyx
→ "What the Next Million Inference Engineers Will Need" with @philipkiely
→ Field notes from token town with @sarahmsachs moderated by our very own @rajaraman_rr

Last few spots, grab yours! https://t.co/V8BR5tGdW2

[Srinath Sankar @Checksum]

My new AI model is deemed too dangerous to be released publicly.

[Srinath Sankar @Checksum]

Can’t help but wonder if the next generation of coding models will be dumber as they continue to be trained on AI slop on GitHub and elsewhere. Does the quality of output drop as the train -> generate -> train cycles increase?

[Srinath Sankar @Checksum]

It’s increasingly difficult and frustrating to read LLM generated docs. There is something about the combination of emojis and unnecessarily verbose language that is off putting.

Much like code, we are heading towards text that only LLMs can ingest.

[Srinath Sankar @Checksum]

Let’s agree on something - if your “AI” pushes slop directly into a repo without human review, my “AI” can indiscriminately and ruthlessly comment and delete that code. Deal?

[Srinath Sankar @Checksum]

Big tech: here is our addictive, immoral, sometimes illegal, all-encompassing platform to get everyone hooked onto our content from which there is no escape.

Also big tech: blame parents, teachers, governments, policy makers and everyone else for causing societal harm.

[Srinath Sankar @Checksum]

Is it still legal to learn something new without a LLM generated post on LinkedIn?

[Srinath Sankar @Checksum]

Using AI tools should be a conscious choice. They are incredibly powerful at certain class of problems and should be used in that context. Using "AI" to show more ads on Google, or to give half baked, incorrect answers across the Microsoft suite just seems like such a waste.

[Srinath Sankar @Checksum]

Every single human activity is destructive to the environment in some way or the other. The problem with this crazy "AI" rush is that excessive resources, energy and water is being burned by Big Tech as they shove AI down our throats & making it difficult or impossible to opt out

[Srinath Sankar @Checksum]

@mitchellh At the very least, if I use a passkey, let that be a permanent session until I log out. But yea, this is an annoyance across the web where multiple sites/apps prompt for a relogin every day.

[Srinath Sankar @Checksum]

@tsriram Awesome! Congrats 🥳

[Srinath Sankar @Checksum]

Haunting view

https://t.co/8K20TwqQHV

[Srinath Sankar @Checksum]

Having opted out of iOS 18 for a whole year, I’m dreading upgrading to 26 to get ongoing security updates. And how Apple managed to screw up the photos app is beyond me.

[Srinath Sankar @Checksum]

Are we now pitting LLMs against each other to get some sort of validation for *our* POV!?

[Srinath Sankar @Checksum]

https://t.co/SqSMI2C0Db

[Srinath Sankar @Checksum]

I think of “Help help I’m being repressed!” at least twice a week.

https://t.co/JtPwfMYhEN

[Srinath Sankar @Checksum]

@krishashok It’s quite jarring when things happen abroad *without* all the red tape. Like what to you mean I don’t have to take a million copies of attested documents, photos, cancelled cheque etc to prove that I’m me!