Me after reading the third consecutive paper that got into an A* conference by implementing an idea I had thought of implementing in November but didn't because the math seemed too difficult
@buccocapital ChatGPT has a bit of an edge in the second category in my experience because model capability/world knowledge matters more than access to available sites and google AU mode seems to be using a pretty weak model
@buccocapital There's a nuance here imo, search has split into two categories, there's your traditional "targeted search", where you know exactly what you're looking for vs the "exploratory search" where you want to maybe aggregate info across multiple sources (1/2)
@arpit_bhayani I feel a corollary to this is that, making sure the signals you send out are not perceived as noise is also an equally important skill to develop
@Samhanknr@championswimmer Heat dissipation still a problem in the desert, even more because water is scarce. Still seems more sensible than space as far as incan tell though
@championswimmer Do we actually have any hard data on how LLMs themselves fare on AI Gen docs ? Atleast in my personal experience of using docs as a sort of "team brain" for the agents, I've noticed it can backfire horribly at times
I think the real tragedy is that the "speed at all costs" approach, in the long term, doesn't really make you that much faster either. All the time you save by ignoring reliability right now, you have to spend fixing broken things later, possibly making you slower!(1/2)
First, the cloud gave us abstractions that made it easier to forget failure. Infrastructure became someone else's problem, and many systems were designed around happy paths with reliability added later ... managed services remain up ...
Now, it is an AI speed-at-all-costs culture that is forcing us to ship code that just works 80% of the time, because it is okay if things fail 20% of the time, but we need to ship faster.
Software was supposed to be predictable. Somewhere along the way, we started treating reliability as a tradeoff, and it is now becoming the norm. I do hope we come back to building software that people can depend on and something we are proud of.
In an early meeting at Facebook (c. 2007), when I was describing the goals of Facebook Platform (an area I oversaw) Bill Gates yelled at me/us.
His quote has stuck with me to this day:
“This isn’t a platform. A platform is where the collective sum of revenues of the participants exceeds those of the platform itself.”
Ladies and gentlemen, I present to you the tokenmaxxing circle jerk.
@championswimmer Don't you think there will be a difference in bench marks that deal with one-shotting tasks (even relatively straight forward ones) vs ones that involve back and forth in a session? IME claude is better at understanding me as a session goes on than say minimax or kimi models
@championswimmer@DFaithG I'm gonna sound like a purist, but what's more likely is "things are broken & shitty but they just barely work & people just lower their standards bit by bit until everything is enshittifed", feel like this has already happened in software to an extent but AI might accelerate it
It's true that no customer really cares about the quality of your code or your attachment to the craft. It's also true that you yourself should still care about those things. Never lose your love for what you do anon, even AI slop can be architected to be beautiful
I'm fascinated by this level of existential crisis developers seem to be going through.
The uncomfortable truth is nobody needs you to be an artisan coder.
Nobody cares about how you coded your app, or whether you feel an emotional attachment to your craft.
You were always code monkey with a high enough salary to believe that your individualist craftsmanship matters to anyone.
It doesn't matter to anyone but you. Not your employer, not your customer. Nobody cares about how you made the product. Nobody cares about your attachment to your process.
You're experiencing the same as countless other artisans have experienced in the last century.
I'm happy for you. You were starting to believe that you're a demigod amongst mortals.
You're not. A machine is better than you.
Now you're free.
I went on an 'arranged' meet up yesterday. We met at the Kind Roastery in JP Nagar. Out of 90 mins, we talked for 85 mins about AI, clawdbot, agentic AI, agile teams, feature anchoring, how pull requests is for low trust teams, the fact that the founder of Linux is also the founder of git, that he went to the same uni as this guy in Helsinki, his dad pursuing a statistics phd after retirement, him using AI to generate code in pascal for something his dad wanted to solve for, etc. Towards the end at about the 85th minute, when we had to call for the check, he was like "I know this wasn't the purpose of our meet up and I'm sorry if I got carried away. Just really quickly, what are your values and what exactly are you looking for" 😂😂 tbf I LOVED the conversation. I love talking to people who are so switched on and passionate and read a lot, and it really shows!
@championswimmer I know you talked about the ad value shifting up, but I don't think that's a good analogy. People will still display ads in the same places, substack and mailing lists will still be valueable, it's just that SEO tricks will change to target AI aggregators rather than human reader
@championswimmer If we go with the email example and if AI really is reading my emails and organising user's day, isn't there a lot of value in ads that jailbreak the AI, so that it puts my ads in the users shopping list for eg.?
Out of all the major cultural issues, one root issue I see in Indosphere business culture versus East Asia is this concept of “how much are you making from this trade”: what you owe me is a function of how much you make, regardless of what it would cost me.
East Asian business culture tends to focus on what they can make and then pass it down the chain. If I make a widget and it costs $10, you take it and price it however you want. My MOQ is fixed. My price is fixed. There’s some haggling on smaller details, but it is what it is, and I’m not interested in wasting time arguing.
Indians are often very interested in how much you’re making and how much they can get from that, and they price their value-add as a percentage of your earnings. This is also a root cause of a massive bribe culture: “You’re making 200, you can’t even pay 20?”
Equity without equivalent risk, basically. This creates a lot of friction and dishonesty in transactions. Everything becomes personal.
Interestingly, bribing in East Asia though largely non-existent in most places is still typically based on absolute numbers rather than a percentage.
This shows up in other variations too, like being overly interested in gossip on how much someone else is making - or who’s earning how much - while doing nothing to improve your own life. Jealousy etc is downstream of that culturally.
@AMR1T1@adityajakki@swarajk_ Also, from what I've heard the deflation, say in food, is driven by oversupply, not falling demand. If we can work out trade deals in the coming days that establish new markets to absorb the supply, we would see some amazing growth in that sector too
@Noahpinion Yeah but they can reach profitability (domestic and foreign) long term now that they basically driven all the non-chinese companies out of business, can't they? There will still be some consolidation domestically, but overall this seems like a successful state led VC model