Shubham☢️

Building KappaX. Digital adoption platform.

[George Pu @TheGeorgePu]

Mistral just open-sourced a text-to-speech model that beats ElevenLabs.

3 GB of RAM. Runs locally. Free.

The thing people were paying per-word for last year runs on your laptop now.

[Shubham☢️ @kill_mockinbird]

The next wave isn't about better websites. It's about capturing the attention of the AI-first human — someone who still decides, but through a completely different lens.

We're not in the attention economy anymore. We're entering the intention economy.

[Shubham☢️ @kill_mockinbird]

Websites won't exist in future.

We're moving toward a world where AI browses, compares, and decides on our behalf. The static webpage, the landing page, the banner ad — all built to capture human attention.

But what happens when the human stops looking?

[Shubham☢️ @kill_mockinbird]

Hot take - AI app UX is shit.
Users need AI native interfaces.

[Shubham☢️ @kill_mockinbird]

@samruddhi_mokal NANO

[Shubham☢️ @kill_mockinbird]

@tomcrawshaw01 ALGO

[Shubham☢️ @kill_mockinbird]

@aryanXmahajan ENGINE

[Hemant Mohapatra @MohapatraHemant]

As a military kid, I don't speculate publicly on war. Besides a few, absolutely no one has all the information that's both correct and timely. I will however say a few things that I've learned in conversations over the last few days with some of the people who are in the know and not afraid to speak their minds:

1) India's strength is that we are basically too big to fall and our weakness is that we have way more to lose via distractions like border wars. We should not make the mistake US made in the 90s that let China get ahead of them in a majority of upcoming technologies.
2) So how do we detract from others dragging us into conflicts that benefit them more (one could argue pakistan basically got $1B of IMF money from this war!)? Simply become 10x more fearsome - so much so that no one dares to provoke.
3) How do we do that? Increase military CAPEX far more than it is today & focus a LOT of it on modernization. We are using equipment that's a few generations behind SOTA in so many categories. Indigenization is the long term strategy, defense deals & tech transfer is the short term.
4) Where you spend this? Focus CAPEX heavily on air and water defence upgrades - our military spend is 8x Pakistan. Our total military aircraft is roughly 1.5x, naval vessel fleet is 1.5x. Almost across the board (army, navy, airforce ), our personnel strength is ~2x. We should be a lot, lot larger! Modernization is a (1) force multiplier - do way more with less people (2) tactical advantage - maximize impact with the right set of equipment e.g. latest drones, advanced soldier gear, satellite systems, military-grade AI, and more!
5) And finally the "soft stuff" - Focus on information warfare. Control the narrative, set up clear, verifiable sources of information that keep the world up to date. We are not world class at this. This takes effort and skill - we should learn from Israel.

We have a lot to focus on this decade internally - war is an unfortunate consequence of having a border to defend but the cost of war >> cost of deterrence and distraction. Especially when we already know India has no point to prove in terms of superiority of military strength to our neighbours to the left. It should be very clear by now to those who know, we can't be beat. @ShashiTharoor said it best - we want to be left alone and work on our nation. We've got a future to build.

[Shubham☢️ @kill_mockinbird]

@adrian_horning_ crunchbase

[Shubham☢️ @kill_mockinbird]

@MaficoNFT Yo! Invite needed.

[Timothy B. Lee @binarybits]

A lot of bad thinking about AI flows from inappropriately anthropomorphizing AI systems. Examples:

[Arthur Mensch @arthurmensch]

Removed, enjoy !

[Shubham☢️ @kill_mockinbird]

@dvassallo It's all about the momentum. In both cases momentum has to come to zero. (Same speed as formula car at crash during impact) * 10X mass = 10X the force of impact.

So same mechanisms won't work for planes even if they were commercially feasible.

[Shubham☢️ @kill_mockinbird]

@cwolferesearch Is it better than FalconLM?

7B models have same ratings on the leaderboard
https://t.co/aSQwBeTEX6

[AK @_akhaliq]

Cinematic Mindscapes: High-quality Video Reconstruction from Brain Activity

propose Mind-Video that learns spatiotemporal information from continuous fMRI data of the cerebral cortex progressively through masked brain modeling, multimodal contrastive learning with spatiotemporal attention, and co-training with an augmented Stable Diffusion model that incorporates network temporal inflation

paper page: https://t.co/XzbqItFtN1

[Amit Ranjan @amitranjan]

Reckoning time for Indian #edtech ??

Chegg - a publicly listed US edtech company lost 40% stock value yday over risks to its future from ChatGPT!

Chegg is into homework assistance & online tutoring - this is a big part of edtech in India! https://t.co/YuX1YxND6c

[typedfemale @typedfemale]

"Anthropic has been heavily focused on research for the first year and a half of its existence, but we have been convinced of the necessity of commercialization, which we fully committed to in September [2022]" is an odd way of saying "we now realize openai was right" 🥴

[Andrej Karpathy @karpathy]

LLM speak 🙂:
- You didn't find some material boring. It had low quality tokens.
- You didn't describe a task to someone. You prompted them zero-shot.
- You didn't say something non-sensical. You sampled at a high temperature.
- The person is not bad/evil, they are unaligned.
- The person is not based. They are just letting you access their base model.
- You’re not learning something new. You’re finetuning.
- It's not confusing. It is perplexing.

This your few-shot prompt to generate more samples.

[Patrick Casey @restoreorderusa]

Brutal