figassis

This is where the absolutism of let agents to 100% of the work fails. You get adversarial agents pulling all reverences into a table, they might miss some, so run this a few times.

Then have another set of agents, with skills like web browsing (to verify that links actually exist, maybe that references and abstracts actually match, etc), have one engineer (or agent) write a small script to help with this (just make sure you test it, and a bit).

So your work is not verified until your references table is 90% green checkmarks, maybe with uncertainty figures.

A human can then verify the ones with under 90% certainty.

This alone gets you a long way there. Does not costs the millions they're being paid.

It's quite interesting that these companies marketed themselves as them best of the best in excellence, accept no mistakes. I can imagine the countless keynotes and books about this. Or the sales pitches.

Has always been a lie, they just understood how to hide it. Today they don't, and it's embarrassing.

So I’m wondering that the end goal and usefulness here is. Is the premise that AI co authored = slop and therefore is bad? And if this succeeds we will start discriminating against co authored code or highly AI co authored code?

As an example, a lot of my new code is AI co authored, but my AI assisted code is probably better than your AI assisted code (subjective I know).

So this could likely add more value if it were scoring quality, architectural soundness, bug/rework ratio, etc. and then you’d be building codescene alternatives.

You've made my point for me. He will never need to spend the trillion personally, it's humanly not possible, but it was already not possible to spend his previous wealth, and if he had started selling that it would crater, but now, he can sell much more of it before he sees the effect, and this goes on. This is what infinite wealth means, you can continually spend more that you could yesterday. Money is money is money.

Try playwright? Works for me

We agreed to its value, and it can be spent and in your limited life span it translates to a drastically different life experience. So, fake money is still money. Not making any judgements on the topic, but waving this truth away doesn’t work.

Sometimes I see these conclusions and I wonder what we’re doing at these research centers because this seemed obvious to be when I was still a teenager.

We are not more wealthy, because the earth has the same resources it had before. We didn’t move to a bigger planet that can now easily feed everyone. If you can extract resources cheaper and faster, it doesn’t mean you have more resources it just means you will hit the limits faster.

What we have done is map those resources to some global ledger and then applied a multiplier to it and moved around who owns most of it. So now you have a class of humans that are fewer and own more of the same amount of resources and use the multiplier to claim we now have more so don’t complain because look at this awesome hockey stick line.

What are these people doing at these labs?

It's a strange thing that as humans, we sleepwalk into every crisis, never agreeing on anything, and then when we're there, we also never agree on the causes. When we ge too the point where we can no longer "engineer" or "science" anything we will spend the next decade arguing that the issue was not really AI, or that if it was, it was inevitable and no one (or everyone) was to blame. Rinse, repeat. Yet we're here, today, looking at the bleak future, and taking yet another step forward.

Do we assume society just self regulates. I think it does, but the cost of letting it self regulate is really really high, with lots of suffering. Is it that we find this acceptable when there is a chance we won't be the first to feel the pain?

RN the market is not you. The amount of investment being pushed to AI dwarfs what you collectively spend. So the market now is driven by the IPO dream where you'll hold the bag, jobless.

> the goal seems to be to create as many things as fast as possible, throw them into the world and see which ones gain traction, whether or not they serve a real need

The goal was never to solve a real problem, like we evangelized for decades. That was how it was explained when resources (mainly time, but also money) were scarce and we could not just throw things at walls. Now we can, and you won't see anyone talk about "make something people need".

Things will be low quality until something sticks, and then money will be poured into it. It's not a bad strategy, but my takeaway from this is: there are multiple plausible explanations for the same thing. People have an incentive to not give you the correct one if it helps you compete with them. But they will give you a sensible one. AI won't protect you from this, experience and real knowledge will.

Everyone will wake up in 2028 and realize Anthropic is the new Google. Not OpenAI.

So because I think it will help us, others might do it first, don’t worry, open algos. Isn’t this everyone’s reason to do something controversial? Also looks like the OpenAI route no?

And it’s too soon to have these norms. Employers today are willing to part with them at the hint of the slimmest efficiency gains, you’ll waste time. So I think the correct response today is wait for it to settle. Norms will form on their own.

It seems this is getting downvoted, no promotion. Understood, thanks for the reminder.

I may have a solution for you and could use some beta testers. Please reach out to me if you don't mind. Info on my profile.

It threatens because we aren’t just talking about selling your art. Artists get hired at companies to produce all kinds of work that will now be replaced by AI.

I shared this here before, I think we're trying to over engineer identity. How about decentralized verification?

https://humanidentity.io, https://protocol.humanidentity.io

Disclaimer: I am the author, feedback appreciated

Thank you for this. Honestly, please earn something. You should at least make it optional to buy/donate. I wanted to, couldn't.

Over editing is one of the biggest tells of junior engineers. Often, a very big task is reduced to a 1 line change if you spend the time to understand the core problem.

I feel like this was one of the most valuable skills an engineer could learn, as it protects the integrity of the system by making minimum viable changes. If you need to refactor something, it should be clear that that is the task.

But we're all tokenmaxing now.

I think HN is the crowd that values MVPs. And LLMs are the best tool to quickly materialize an idea. So I think we should judge these submissions on merit and not on our collective rejection of reality. If they succeed I’m sure (or hope) their user facing app won’t remain vibe coded.

What is true and ironic here is SWEs never thought they’d need to unionize.

Maybe I’m need coffee, but I don’t understand what DebtDrone actually does.

The interesting thing is this: no matter who they hire (because certainly they’ve already been hiring the best), will make a bank Mythos proof. It likely won’t even make it regular attacker proof.

The reason banks aren’t being hacked daily is a mix of physical and digital access controls, and processes, as well as the real threat of the FBI knocking on your door. It is not because the systems are hard to hack.

But now, they panic because some company told them their systems were not 100% secure? And how will they solve this? My making it 99.99% secure with a $10B yearly Mythos subscription?

I would just settle for not leaking any more s3 buckets. Start there.

I see this pattern for a long time. Large companies always release the same products at the same cadence, hours or minutes apart. Are they having the same ideas? Spying on each other? Are these products just the natural next step from the previous so as long as you’re not stopping, 10 companies would release the same products?

This is correct, and I avoided it for this reason, did not have the bandwidth to get into any cpp rabbit hole so just used whatever seemed to abstract it away.

Thank you, this is very early stages. Still trying to validate the idea. But yes, the reason there is a sovereign verifier tier is because I am sure governments will want their own rules, and the protocol is meant to be decentralized. So one govt can legislate that they are the exclusive verifier for their country, while another takes a more hand off or hybrid approach.

If this is the case, this can be gamed. People can use stolen documents. Nothing says a person can’t own multiple computers so what happens if someone uses your id in 20 laptops? Will the companies just claim “but the machine said they where old enough?” The law may not have teeth, but will violate privacy.

Something like https://protocol.humanidentity.io (disclaimer: I built it, sorry for the plug) or any other privacy preserving service might work better. A platform can then require that a person verifies age in a privacy preserving way before viewing adult content.

From what I’ve seen, this always ends in some small fine/settlement and “no admission of guilt”. This type of protection is the source of these mishaps.

This is a hard problem. I understand DEI, and I understand anti DEI, and agree with both in that DEI aims to close a gap in disadvantaged groups that were discriminated against in the past to the point that they are demographically cut off from wealth because they can’t gain any escape velocity to allow them to acquire assets that work for them. They’re locked into the jobs the wealthier don’t want, regardless of their skill level. And even getting to that skill level is a challenge because they’re too busy working 3 low paying jobs to attend school.

On the other hand, members of a demographic that indirectly benefitted from discrimination in the past, but were themselves not direct beneficiaries can’t be punished for that. They apply to jobs just the same, compete just the same, and aspire to be successful just the same. And just because they can’t relate to the fact that the other demographic has it harder, doesn’t make them evil.

So where’s the balance? What’s the solution? And companies implemented DEI bc they were pressured to do it, and now they’re pressured in the opposite direction. Will they be sued again in a few decades for the other kind of discrimination of “we only hire the best, it just so happens that all the best are of X demographic”?

This is why you only use virtual cards for subscriptions. I have never had this issue. Even adobe couldn't get a cent over what I was willing to pay. Didn't let me cancel the account, fine. Kill the card. You don't have to beg to these companies.

It's not the feature itself, it's how/what the llm tries to use it for. It uses it to cross any and all architectural boundaries.