jonas_kgomo

Some work is unpaid, just like open-source work is unpaid but gets you experience. Some work is paid. In a perfect world you dont need experience to get a job, we arent in a perfect world

Can we build a different app experience on this open-source tool? Does the license permit

I watched a movie a few days ago and they basically said there are more states in the game of chess than atoms in the universe? https://www.youtube.com/watch?v=xfMQ7hzyFW4

I actually found it quite robinhood for openai to acqhire, bascially this startup was my favourite thing for the past few months, but they were experiencing server overload and other issues on reliability, i think openai taking them under their wing is a good/neutral storyline. I think its net good for science given the opai toolchain

I tried watching this over a couple of months and still remain lost every 5min. There seems to be a certain jargon they use that transcends normal speech, and they are operating in the same linguistic sphere. Does anyone understand any randomly selected 5-20 min of this discussion

This is the most difficult conversation i ever listened to, because each sentence had so much layers that I could not possibly understand every 5 min of the discussion. My questions is, how does one get to understand these kind of conversations, and secondly, where can I find more of them? NB: I guess the word here for high-density should probably be entropy or information content as opposed to density.

This is the most difficult conversation i ever listened to, because each sentence had so much layers that I could not possibly understand every 5 min of the discussion. My questions is, how does one get to understand these kind of conversations, and secondly, where can I find more of them?

NB: I guess the word here for high-density should probably be entropy or information content as opposed to density.

this is basically a direct copy of the startup that got purchased by linktree right?

I have shared a similar post about another app of the same kind. It is called Invisibility (already using it, but concerned)!

https://news.ycombinator.com/item?id=40530354

what examples are you considering here, bioweapons?

Our key insights are: 1. You don’t have time to do everything you want, you must make tough choices

2. Set a priority for each day, and plan your day strategically around it

3. By reflecting on the week gone by, you can reset yourself for the week ahead

4. Reflecting on your beliefs and knowledge(epistemic housekeeping) can be essential to removing overestimation and bias in your planning

very interesting, the uLog page looks great. i will try this

Foundation Models Taskforce | Advise | Expression of Interest | UK

The UK government has committed an initial £100M towards its Foundation Models Taskforce, the largest amount directed by any nation towards AI safety. We want to find people with diverse skills and backgrounds to work in or with the Taskforce, to catalytically advance AI safety this year with a global impact. Apply: https://docs.google.com/forms/d/e/1FAIpQLSeOV3FPChEstFy7CcuH...

Andromeda Cluster | Grant | Remote | Compute (10 exaflops*)

Available for experiments, training runs, and inference. No minimum duration and superb pricing. Big enough to train llama 65B in ~10 days. For use by startup investments of Nat Friedman and Daniel Gross. Apply: https://andromedacluster.com/

Emergent Ventures Grant | Fellowship | Remote | micro-grant

WHO: We want to jumpstart high-reward ideas—moonshots in many cases—that advance prosperity, opportunity, liberty, and well-being. We welcome the unusual and the unorthodox. We encourage applications from all ages and all parts of the world. Apply: https://mercatuscenter.formstack.com/forms/emergent_ventures

I remember seeing Boston Dynamics videos and thinking , this is probably in the +10 years timeline. However there has been increasing papers [0] showing how robots can be grounded with LLMs. As artificial intelligence (AI) continues to develop, it is becoming increasingly possible to create robots that are capable of autonomous action. This raises a number of safety concerns, as robots could potentially be used to harm people or property.

One of the main concerns is that robots could be used to carry out malicious acts. For example, a robot could be programmed to attack people, steal property. In addition, robots could be hacked and used by malicious actors for their own purposes. In addition, robots could be poorly designed or manufactured, and could malfunction in unexpected ways. Specifically in this demo, this robot can be prompted to do things unforeseen.

Here are some specific questions that need to be addressed:

How can we prevent robots from being used to carry out malicious acts? How can we prevent robots from causing accidents? How can we ensure that robots are designed and used in a safe and responsible manner?

[0]https://ai.googleblog.com/2022/08/towards-helpful-robots-gro... [1] https://twitter.com/MetaAI/status/1670484372309573632

It's good to note that this product has an open-source version https://react.email/docs/introduction

I have seen DAIR members perform ad hominem attacks on qualified researchers and first movers in AI Safety (eg EA community, Yoshua Bengio, etc), they divert the discussion away from x-risks and focus on attacking individuals. For instance, look at the reaction of AI pioneers expressing interest in studying ai safety. This people seem to be focusing way more energy on attacks and discreding than making actual progress that addresses the case of ai safety. I do not understand diverting discussions about existential harm with diversity/patriachy, they seem to be separate discussions. Whereas they seem to attack anyone working on AI Safety who is a white man.

https://twitter.com/Abebab/status/1667150619029676033

DAIR: https://twitter.com/timnitGebru/status/1667251985844666369

The field of AI Alignment seems to have a long list of hard problems that could be publicly researched openly. Inspired by The Polymath Project, Sir Tim Gowers & Michael Nielsen.

I am wondering where this could go, write a research prompt(leaf) and another person continues? How does this look at scale?

Page: https://progress-studies.org/polymath Playground: https://progress-studies.org/polyAI Polymath project: https://en.wikipedia.org/wiki/Polymath_Project

Another expert said: The situation doesn't sound globally dangerous - but you don't want to get a cholera, polio or measles outbreak there right now. Hard to judge risk: more likely stuff gets destroyed than spreads. https://twitter.com/anderssandberg/status/165087086024744140...

For context here is a link: https://www.cbsnews.com/news/sudan-virus-risk-world-health-o...

> Fighters have occupied a national public laboratory in Sudan holding samples of diseases including polio and measles, creating an "extremely, extremely dangerous" situation, the World Health Organization warned Tuesday. Fighters "kicked out all the technicians from the lab... which is completely under the control of one of the fighting parties as a military base," said Nima Saeed Abid, the WHO's representative in Sudan

Thank you, i will update this with a video. Given the feedback, it makes sense to let the issues be public as well

Basically, github is used for authentication here, then a user is able to add issues and see other issues. The idea is to have a public/private error documentation and explanation page. That is, how/where the error occured and how a person solved it. It can be useful for teams onboarding new people facing similar integration errors, or just for personal logging for future reference

No information except for user-name and profile picture is used here. Its using Github OAuth and not Github Apps.

Thanks, I have sent you an email. Let me know if you got it!

Very interesting project. i was wondering about the cost margins of running this. and how does the workflow from gpt3 react code update to ui

Call for algorithmic energy efficiency initiative: it’s estimated that GPT-3 consumed 936 MWh, which would be sufficient to power ~100000 European households for a day. Ethereum has 99% efficiency by changing their algorithm. Is there a way to design algorithms and hardware to be energy efficient. Can we have energy complexity just like we have space and time complexity?

For comparison annualized energy consumption (TWh) of technologies : Bitcoin 100, YouTube 244, PoS Ethereum 0.0026.

https://ethereum.org/en/energy-consumption/

Beyond racism some of these things happen a lot in startups. As a black person as well, i might have experienced this in a much lower stakes, for example, hired at a startup.

https://news.ycombinator.com/item?id=22091777 The behaviour sounds totally unacceptable. "Is there a way to do this without making it painfully apparent to him that he's being diluted to 10%?" Zuck.

Sampling in hip hop is about taking 1 or 2 audio sources mapping them into some new audio, SD was trained on 2.3 billion data points. In a diffusion model you put noise on an input image, until its recognizeable. For a song this would mean, making it unrecognizable by making it literally sound like noise, and then try to denoise it, this process wouldn't be very helpful i assume, but might be a good experiment. Open AI's Jukebox uses an autoenconder method with autoregressive transformers, given this architecture, i would assume you can theoretically do stable diffusion architecture on audio, to generate new audio. https://openai.com/blog/jukebox/

I was at Palintir, and one of the things I learned from visiting is that they thrive in times of war, pandemics and recessions. Why aren't companies designed to be antifragile like this