jimsimmons

Everyone knows Google is an ad tech company. Real question is if they can diversify long term to be something beyond. Cloud is a great way to achieve this

Google but allowing incremental queries. Incremental!= conversational a la ChatGPT. So I can tell Google to ignore an interpretation of my query so I can drill down to the exact thing. I don’t like writing full sentences or questions to ChatGPT especially when it doesn’t have auto complete.

Generative content sprinkled here and there but majority of results still being genuine human created content.

I don’t see MSFT being a player here. Their ML talent is shallow, their search team is tiny compared to Google and a partnership with a startup won’t change that

Sorry this is totally unsubstantiated fantasy. You’re not even wrong. I don’t know how to point that out here

At least on HN can we stop parroting this supposed open book strategy from 5D chess Nadella?

Bing has failed to take away meaningful marketshare from Google by MSFT’s on admission

https://www.theinformation.com/articles/microsoft-eyes-firef...

They went for the Google++ model. That isn’t enough to cause a switch. You need Google x 10 or 100 to unseat the beast.

I’m very glad they tried though. Gives us hackers hope that building a Google quality index outside of Google is not unimaginable

They over estimated the market on the table —- in tech circle people want to end Google and are willing to pay. For the average user, they have no clue how the ad business works or harms society

Appreciate the honest retrospective.

This combined with the reports that Bing failed to cut into Google’s market show that Google is safe for a while.

I wonder if I’ll see disruption of Google’s core business in my lifetime. I really want it to happen but it’s a mighty quest

How does it work for coding? Are we really doing trial by trial analysis of code these days

It isn’t seen an condoning suicide. GPs point is more that his eventual suicide makes him less of an authority here. Bit like a relapsing addict trying to justify his drug use

Badami is highest quality followed by Raspuri imo. Former for milkshakes and later for raw eating. Everything else is noise ;)

This. It’s like people in the west have never eaten real mangoes and only know of it via artificial flavour.

I’ll also mention Raspuri which is my favourite kind. As the name suggests it’s very juicy

Layoffs and removing staplers are management’s way of saying good times are over. They want to stop rest and vest culture. It has nothing to do with actual value of those employees or the financial necessity

Somehow all the non programmers are in full force telling actual coders how their job is changing

You are the author so can change the title?

You can’t have the cake and eat it too

Proto AGI GPT4 can’t faithfully translate Python to Julia?!

No classes, weak typing despite having multiple dispatch, 1-based indexing, no real DL libraries with comparable baseline to pytorch/weight portability

There are no reasons to not use it if you like gesturing from a distance

We’re not disagreeing?

My only point is that they are not necessarily a bleeding edge tech company. I agree they have a huge, potentially unassailable, moat.

This announcement is not about AlphaGo or GPT level tech reaching the military. It will be perceived so and the CEO is happy to tell that white lie

Their value add is in human services and integration, sort of similar to consulting, and not technology. So using “AI” and dazzling government contracts is misleading imo

Palantir jumping on the AI bandwagon. They have a IT systems solution, nothing more.

This is why no one has a simple explanation of what their technology is ever. Also why they’ve not scaled like conventional tech companies. Just like someone like Infosys, they need forward deployed engineers for each customer to build out their information management and aggregation system.

26000

What if I care about like 4 archs: x86/64, arm64 and PTX

Because I want to be self hosted

(tbc I was forthcoming with my arrogance, like it genuinely didn’t make sense to me)

Why would constant propagation need a PhD level proficiency to implement? It’s basically running part of your code and no different from parsing.

Again not trying to trivialize or be rude. I wish this was something I could ask ChatGPT and trust its outputs but you know I can’t

Let’s say I’m targeting x86 or PTX or some standard subo instruction set. I can do vanilla codegen but it is slow af

Question to pros: Things like constant propagation, loop unrolling, TCO, memory reuse etc., seem simple in theory. Yet why do languages end up depending on LLVM and not implement these algorithms on their own and avoid a monstrous dependency?

Like what exactly is the implementation of the top 20 optimizations? It can’t be that high right..

It doesn’t say exclusively for the beta period

Source?

Confirmation here:

https://twitter.com/AnthropicAI/status/1656743460769259521?s...

Let’s not let facts interfere with psyops

Wow. Is there anything equivalent for men? I can’t imagine spending 1K a month total. How could you even spend it