Google Colony Ship

@CSI_Starbase @JordanDrouin It's both: Elon is disgusting people and the current phase of Starship development is boring them. My question is do you think the former effect will still be strong when the latter effect flips with crewed missions?

@JordanDrouin Guess you didn’t read the replies.

It’s very disheartening to see how rapidly the amount of people interested in the Starship program has taken a massive nose dive. Seems to be a 50% drop if not more within the last two months. YT analytics unfortunately spell this out quite clearly. Nothing is going to change my level of passion, but I can’t do anything about how everyone else feels 😔

@Replit Thank you for logs searching!

We also added CPU & RAM monitoring to Deployments. Now, you can easily access system information about your Deployment machine right in the workspace.

Today, we are releasing Object Storage, an improved logs experience, and CPU + RAM monitoring for Deployments. 🧵

@JD_2020 @gdb @OpenAI @miramurati @WeGPTai Perhaps you can select the subset of the SWE-Bench tasks suitable for WebGPT and compare its performance to auto-code-rover (above) and github.com/princeton-nlp/… (which do slightly better and worse than Devin's claimed score by the way)?

I think the two biggest things holding me back from doing these two features right now are resource related. 1. To do a proper SWE bench, I feel like we need a general purpose playground and right now these playgrounds are just p5js environments, 2. To add a GitHub issues endpoint, I think again, a general virtual code execution environment for testing and verifying contributions is necessary. Though possibly not. This was a great idea and I’ll think more about this.

Using WebGPT🤖 to build & deploy a playable version of FLAPPY BIRD… ☝️ directly from chat ✌️ in just SIX prompts 🤘 with absolutely NO CODE (Links in 🧵) @gdb @OpenAI @miramurati

@ry_serene @SergioRocks Tell him it should probably be closer to 75%

@SergioRocks Third option- Do nothing. He's doing his job as it should be done.

A Software Engineer in my team told me recently: - "80% of my code is written by Github Copilot or ChatGPT" What should I do?

@KlausJennings @lhfang @Noahpinion @RobGeorge Are you talking about Noah or Ian?

@lhfang @Noahpinion @RobGeorge C'mon Lee this dude's the biggest race communist out there except this one time when he feels ((his people)) are threatened he feels different. Look at his tweets during the BLM riots they were atrocious

These are largely pro-war protests, whose goal is "decolonization" via the invasion and destruction of Israel. Whether you agree or disagree with that goal, you should be honest about what the goal is!

@MoodyBlueshirt @lhfang @Noahpinion @RobGeorge Bremmer used the words on the WaPo caption. Why is everyone piling on him instead of @washingtonpost?

@lhfang @Noahpinion @RobGeorge It's a shallow observation, but an accurate one. Which just goes to show how blatant Bremmer's lie was.

@DanJakaitis @ianbremmer Why is everyone piling on Bremmer for this when he was merely using the words from the WaPo caption? Is he on the editorial board?

map of recent antiwar protests across the usa via wapo

@Robotbeat @rocketrepreneur To do it right you need to go explicitly glomar on some batshit explanation to nip conspiricy theory proliferation off at the bud. E.g., "we neither confirm nor deny that these modules will be used for the 55th Armed Space Rangers Division" or some such.

@rocketrepreneur Well, it is defense related.

But why tho? Not answered. “The company did not elaborate on the specific work it will be doing under the SBIR or how its module technologies would be used for tactically responsive space.”

@GyorgyKriza @dorfman_p @skdh Where do you put the waste? How much does the refined fuel chain security cost to keep bombs out of the hands of terrorists?

@dorfman_p @skdh That is because of the high construction price of nuclear power plants. But once they are running, they are unbeatable.

News are erupting in Germany that the parliament's decision to go through with the nuclear phase-out was partly based on false information that seems to have been deliberately falsified. cicero.de/innenpolitik/r…

@Davejones0305 Ask him whether he thinks repealing austerity would work.

Just been on my local community Web site stating how bad the economy is and how it come to be, via tory austerity and guess what!? Yes some tory voting bovine cvnt jumps on, well I ain't concerned about the past, it's about maintaining a buoyant economy now?! Wtaf 🙄 Div

@sperformance @Grady_Booch Nothing AI can't fix with a more modern flair.

@alexalbert__ How about sending an announcement of such contests to customers by email in advance?

Congrats again to the winners! We are looking to host more events/programs like this in the future! Let me know what we should do next👇

Our first Build with Claude contest was a success! We received tons of great submissions from @AnthropicAI devs. Here are the 5 winning projects (in no particular order)🧵

@Robotbeat But isn't that the actual position of both sides? Perhaps superpowers forcing them into negotiations is more important than nuance?

So, like, stuff like this, the harassment of Jewish students, and the fawning over October 7th, the ignoring of the hostages, etc makes me unable to full support these protests. I KNOW activists usually have a hard time pursuing nuanced issues, and this situation REQUIRES NUANCE.

Me: "Partner, you can now buy a robot flamethrower dog for less than 10K. Technology advancing, yknow." Partner: "Huh. How's cancer research going?"

@malcolm_reavell The effect occurs under socialism and all forms of pre-communism which have ever existed. Changing the rules isn't necessary, just the progressivity of taxes and transfers.

@GoogleColonizer Upward transfer of wealth is only inevitable in a capitalist system. It's a feature of the system. This is one reason why a UBI is not a good idea. Giving everyone free money will just feed into and inflate an already badly skewed distribution.

Here's a short quiz: Are these statements true or false? Note your answers for later.

@AukeHoekstra The weird thing is she doesn't acknowledge the existence of plug-in hybrids.

Maybe we can crowdsource debunking this video? I will try to do a thread on it tomorrow but in the meantime I would love if my followers watch it reply with the main points where they think Sabine goes wrong. (By the way: she was already against electric cars in the past.)

@skdh Have you heard of plug-in hybrids?

@8teAPi LLM progress will be incremental, but support modules surrounding them won't. Working memory, goal-directed self-critique, enhanced RAG, will cause agent capabilities to increase several fold in the next few months. We are seeing this with many superior Devin competitors emerging

Zuck on Dwarkesh TLDR: AI winter is here. Zuck is a realist, and believes progress will be incremental from here on. No AGI for you in 2025. 1) Zuck is essentially an real world growth pessimist. He thinks the bottlenecks start appearing soon for energy and they will be take decades to resolve. AI growth will thus be gated on real world constraints. > "I actually think before we hit that, you're going to run into energy constraints. I don't think anyone's built a gigawatt single training cluster yet. You run into these things that just end up being slower in the world." > "I just think that there's all these physical constraints that make that unlikely to happen. I just don't really see that playing out. I think we'll have time to acclimate a bit." 2) Zuck would stop open sourcing if the model is the product > "Maybe the model ends up being more of the product itself. I think it's a trickier economic calculation then, whether you open source that." 3) Believes they will be able to move from Nvidia GPUs to custom silicon soon. > "When we were able to move that to our own silicon, we're now able to use the more expensive NVIDIA GPUs only for training. At some point we will hopefully have silicon ourselves that we can be using for at first training some of the simpler things, then eventually training these really large models." Opinion Overall, I was surprised by how negative the interview was. A) Energy - Zuck is pessimistic about the real world growth necessary to support the increase in compute. Meanwhile the raw compute per unit energy has doubled every 2 years for the last decade. Jensen also is aware of this, and it beggars belief that he does not think of paths forward where he has to continue this ramp. Over at x.ai, the launch announcement read "At xAI, we have made maximizing useful compute per watt the key focus of our efforts." So energy efficiency, algorithmic and otherwise are obvious areas where firms will be focused on. Zuck meanwhile is planning to move of Nvidia chips soon, and basically believing that the value in AI available to having a share of the most advanced compute clusters will declline. B) AGI Negative Zuck fundamentally > does not believe the model, the AI itself, will be the product. > It is the context, the network graph of friendships per user, the moderation, the memory, the infrastructure that is the product. > Allows him to freely release open source models, because he has all of the rest of the pieces of user facing scaffolding already done. An actual AGI > where the a small model learns and accompanies the user for long periods > while maintaining its own state > with a constitution of what it can or cannot do > rather than frequent updates from a central server > would be detrimental to Meta’s business, > would cause a re-evaluation of what they are doing C) Summary Zuck > essentially settled into the trap of believing in incrementalism > Advised by the smartest people in the world. > Technically competent > But he does not believe in states of the world where a 100x improvement from GPT-4 are possible, or that AGI is possible within a short timeframe. > But he's also… he’s not raising capital. The three people who are raising capital: Sam Altman, Elon Musk and Dario Amodei, are all on record expecting dramatic increases in capability. They could be hyping because they need higher valuations. I don’t know. And so again we wait for GPT-5. If it is 10-100x as good as GPT-4, the current benchmarks won’t even work (how does one measure 100x as good on an MMLU scale of 1-100 ?). If the models deliver value, and deliver large amounts of value exceeding manifold the capital deployed to develop them, then progress will continue. If not, not. For me, the most exciting part of all of this is.. 🍿🍿🍿 drama. You get to see who is bluffing, with billions of dollars on the line, on a fairly short timeframe. And I, for the most part, am in it for sheer entertainment value in spectating potentially the greatest game mankind has ever played. Happy Friday 🥂 I’d like to congratulate Dwarkesh for an excellent interview… and for not great ad reads as well 😜 The above was excerpted from my daily AI newsletter, full edition goes out Monday. Sign up below!

@joshwhiton Mimicry?

What do you call it when AI does a thing better than us, yet people believe AI is doing a fake version of that thing? Like the ability is counterfeit, but the counterfeit works better than the original. We’re gonna need a word for that.