dpiers

You should see the spaghetti code that humans working 70 hour weeks for months on end to meet a launch date have pushed out.

You can be in the driver seat while delegating implementation to agents. You can use agents to generate documentation and refactor architectures. You can ask an agent to explain a feature and then work with it to plan how to extend it.

You don’t have to type the code to be the architect.

"No more hand-written code"

Isn't this the new norm for every modern tech company?

Claude Code can help me write and review code faster and better than a human possibly could. I spend my time drafting requirements, steering agents and reviewing diffs instead of writing code manually.

AI is changing knowledge work. Dressing that up with ragebait (game dev, Disney, layoffs) doesn't change that or the fact that the video game industry has always been brutally fickle and hit-driven. The Epic layoffs earlier this year had far more to do with the decline of Fortnite's popularity than AI.

Companies used to *need* to go public in order to either raise capital or because they were going to have to start publicly reporting anyways due to the number of shareholders.

The JOBS Act and proliferation of double-vesting trigger RSUs effectively negated these forcing functions and 'going public' went from a necessary growing pain to a burden and distraction.

Tristan Buckmaster’s post cited extensive use of LLMs in the process of his collaboration with Levent:

“We used several LLMs throughout: Anthropic’s Claude, OpenAI’s Codex, especially with GPT-5.6 Sol and, more recently, Astra.

The latter was only used for writeups and auditing our arguments. For most of the past year progress was slow. We worked through the literature and upgraded various preliminary results, up to obtaining finite time blow up for the Incompressible Porous Media equation (with smooth forcing).

This was until about a month ago, when we had real progress: on August 15th, we obtained the blow up results, with smooth forcing, for both Boussinesq and Euler. I can say the first LLM generated proof Levent sent me was the most horrendous I have ever read; we verified it on Lean on August 22nd. Since this point, we have been working around the clock to understand this proof and turn it into something readable.”

The OpenAI research post states they began training GPT-6 internally on August 28th, and that user chats are used to train models.

“We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed in order to solve this problem. While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models .”

For an incredibly niche topic like this, I believe it’s extremely likely that Buckmaster/Levant’s work would influence the direction of OpenAI’s agents’ work even as a de-identified drop in the overall bucket of training data.

We already live in a world where parents decide whether or not a child with Down Syndrome will be born.

60-90% of prenatal diagnoses in the US result in an elective termination. The number is nearly 100% in Iceland and some other Nordic countries. Unlike autism or ADHD, we have a very clear understanding of exactly what causes Down Syndrome and now potentially how to correct it. A treatment like this is no different from correcting a congenital heart defect - it gives a baby a chance at normal, healthy development.

You’re neglecting the fact that each one of those businesses had a plan, they just pivoted to more successful plans.

The title is a lie - the original statement was that it was closer to “a small game engine” than a basic text UI.

CC has a complex scene rendering pipeline with a 16ms/frame budget. That is much closer to a game engine than a print statement.

Languages are thought encodings.

Most people can only encode/decode a single language but an LLM can move between them fluidly.

Amazon has surpassed Newegg as my online parts provider of choice.

It's not that things can't go wrong, it's just that they are much better at handling it than Newegg these days.

We are in the process of abandoning ChatGPT Enterprise because Google is now bundling Gemini access for free with Google Workspace.

If OpenAI is not safe from anticompetitive practices then no YC startup is.

This explains a lot about Ork Mekboyz.

Well there was that one time Germany owed everyone a lot of money after the Treaty of Versailles and it kind of directly led to Hitler’s rise to power and another World War… makes sense for them to be a bit more conservative than other countries.

It seems you are under the impression that the venture capital firm cofounder by Marc Andreessen and Ben Horowitz was created by a single individual.

If you can’t get basic facts correct no one is going to take your other claims seriously. A16Z has a diverse portfolio of investments and are as famous as they are because of how successful their companies have been.

I don’t disagree. Crypto is neat but has no useful (legal) real world applications.

Oh, please. Bitcoin is a protocol with no more inherent value than SMTP.

Aside from being the Catholic cryptocurrency, it has been surpassed technologically in every way by newer and better designed protocols. Why would anyone who doesn't already own Bitcoin choose to use it for anything over, ex. Ethereum?

Bitcoin maxis aren't technologists; they're grifters trying to convince you their magic beans are the most special magic beans.

I’m creative but have never been adept at sharing visual ideas in my head with the rest of the world. I’ve found genAI image tools empowering because for the first time I can turn an idea into my head into something other people can see and enjoy. I’ve used it for everything from portaits and memes to recreating scenes from my memories that I don’t have photos of to help me remember.

I use credit cards for all of my spending. One gives me 4% back on fuel, another 3x points on travel and dining, etc. Paying with cash/debit is less economical because you forgo these rebates and incentives.

I also pay my cards in full each month so there are no interest charges, just a few with annual fees that are more than covered by the card benefits.

Is this news? I'm genuinely asking - TLO and other skip tracing tools have been around for a while, and have been offering license plate location data for years.

The ACLU issued a report on how ALPR devices are used to track people in July 2013 (1).

Flock Safety (2) was founded in 2017 and has raised $381M from name-brand VCs like Tiger Global and A16Z. The ACLU raised concerns about them a year ago in a report. (3)

No one has been keeping it secret.

1: https://www.aclu.org/documents/you-are-being-tracked-how-lic...

2: https://www.crunchbase.com/organization/flock-safety

3: https://www.aclu.org/report/fast-growing-company-flock-build...

This story by itself might sound like sour grapes, but in combination with other reports about the CEO(1), I suspect bad news ahead for Stability's investors.

It seems like a narcissist with a very 'dynamic' relationship with the truth chased the AI hype train and ended up with a bunch of money and attention due to a stupid VC. He pissed off the teams that invented Latent Diffusion and his partners at RunwayML in the process, and now it seems like his cofounder as well. What value is there in a company that's only famous because they spent $600k in VC money training an open source model on AWS?

1: https://www.forbes.com/sites/kenrickcai/2023/06/04/stable-di...

I had a former employer call once and ask me to sign a form confirming that I had opted not to exercise the stock options worth 0.125% of the company when I left.

They were raising a new round and my old grant was a sticking point because the new investors weren't content with the company not having a record of my exercise.

I signed because it helped them out, cost me nothing and seemed like the right thing to do. Just think it was funny that a VC was so averse to dead equity that they made the company make sure someone who wasn't on the cap table agreed that they weren't on the cap table.

Comcast "forced" me off of mine by upping my rate to $180/mo for the cheapest cable-only package when there were contract deals available with internet and more channels for $110/mo.

No contract rates available without turning in the cablecard. Switched symmetric GB fiber provider for $65/mo and pay for streaming TV during the NHL season.

That’s because SMS has a 1120 bit message capacity and uses one of two encoding mechanisms.

GSM-7 uses 7 bits per character and a correspondingly limited character set that allows for 1120/7 = 160 characters.

UCS-2/UTF-16 uses 16 bits per character for a total of 1120/16 = 70 characters per message.

You’re confusing stupidity with narrow expertise. I know CPAs who don’t file their own taxes because that wasn’t the focus of their niche. I know surgeons who don’t know the difference between a transistor and a resistor.

Plenty of smart people don’t understand how ChatGPT works or what it’s limitations are. A bunch of nerds built the best BS generator in history and marketed it as a super intelligent computer. If you ask it for relevant cases and it spits out a bunch of plausible information, is it really on them to know the tool is just really good at making things up?

Their progress reports pop up on HN pretty frequently, as they're usually pretty technically dense and interesting.

Here's the discussion of the most recent update: https://news.ycombinator.com/item?id=36019485

Except some municipalities with large red light camera revenue streams use bad math to justify unsafely short yellow light times in order to increase ticket revenues.

More details here: https://redlightrobber.com

LoRA: https://arxiv.org/pdf/2106.09685.pdf

LLaMA: https://ai.facebook.com/blog/large-language-model-llama-meta...

Both have been the subjects of numerous HN posts in the last month.

I presume the last one is for ensuring the safety of staff. Domestic abusers can use work schedules to stalk and harass their former partners.

I’m an auditor at a large organization that uses Sourcegraph, and I love it. On one project I had to trace a compliance flow across 14 services written in 4 languages, and Sourcegraph made it easy to navigate everything and document our findings with links to code at specific revisions. I use it to dig through our codebases on a daily basis.

Bring a rechargeable battery with you, plug that into a charger on the wall, and plug the phone into another port of the battery.

Except this is the result of a $20m NSF grant to one of the top CS schools in the country, not a fly-by-night startup looking to profit off a trend.

I’m actually kind of interested in this. I’ve taught myself a decent bit and done my own projects but going through a structured curriculum with a cohort of peers has its benefits.