1801
The OpenAI Keynote
OpenAI’s developer keynote was exciting, both because AI was exciting, and because OpenAI has the potential to be a meaningful consumer tech company.
New
The latest blog posts to take off on Hacker News. 3,091 posts.
1802
Formalising modern research mathematics in real time
(This is a guest post by Bhavik Mehta) On March 16, 2023, a paper by Campos, Griffiths, Morris, and Sahasrabudhe appeared on the arXiv, announcing an exponential improvement to the upper bound on R…
1803
AI and Open Source in 2023
The Highs and Lows: A Year in Review
1804
Telling GPT-4 you're scared or under pressure improves performance
Researchers show LLMs respond with improved performance when prompted with emotional context
1805
How to Think Computationally about AI, the Universe and Everything
In his TED Talk, Stephen Wolfram covers the emergence of space by the application of computational rules to spacetime, gravity and quantum mechanics to AI and LLMs. Computational irreducibility and the ruliad.
1806
When gradient descent is a kernel method
Suppose that we sample a large number N N N of independent random functions f i : R → R f_i \colon \R \to \R f i : R → R from a certain distribution F \Fc F and propose to solve a regression problem by choosing a linear combination f ˉ = ∑ i λ i f i . \bar{f} = \sum_i \lambda_i f_i. f ˉ = i ∑ λ i f i . For…
1807
The convolution empire strikes back
ConvNets Match Vision Transformers at Scale
1808
Now add a walrus: Prompt engineering in DALL-E 3
Last year I wrote about my initial experiments with DALL-E 2, OpenAI’s image generation model. I’ve been having an absurd amount of fun playing with its sequel, DALL-E 3 recently. …
1809
1810
Two Kinds of Techno-Optimism
A few things Marc Andreessen forgot to mention
1811
Embeddings: What they are and why they matter
Embeddings are a really neat trick that often come wrapped in a pile of intimidating jargon. If you can make it through that jargon, they unlock powerful and exciting techniques …
1812
What every developer should know about GPU computing
A primer on GPU architecture and computing
1813
"A Young Lady's Illustrated Primer" Simulated by GPT-4
A short simulation of Neal Stephenson's Illustrated Primer using GPT-4 and DALL·E 3.
1814
Interviews in the Age of AI: Ditch Leetcode – Try Code Reviews Instead
In the emerging age of AI generated code, is leetcode really the best way to evaluate technical candidates (was it ever)?
1815
Causal inference as a blind spot of data scientists
Throughout much of the 20th century, frequentist statistics dominated the field of statistics and scientific research. Frequentist statistics primarily focus on the analysis of data in terms of probabilities and observed frequencies. Causal inference, on the other hand, involves making inferences about…
1816
Multi-modal prompt injection image attacks against GPT-4V
GPT4-V is the new mode of GPT-4 that allows you to upload images as part of your conversations. It’s absolutely brilliant. It also provides a whole new set of vectors …
1817
"Building Machines That Learn and Think Like People", 7 Years Later
In the post about GPT-4V, I mentioned the paper “Building Machines That Learn and Think Like People” (https://arxiv.org/abs/1604.00289) by Brenden Lake, Tomer Ullman, Joshua Tenenbaum, and Samuel Gershman.
1818
Every app that adds AI looks like this
and just like that - poof! - we scammed you into thinking this shit was cool // S1E3
1819
OpenAI is too cheap to beat
Data matters; infrastructure matters more
1820
The AI research job market
There are plenty of jobs, but finding a place where you're happy is as hard as ever.
1821
We’ll call it AI to sell it, machine learning to build it
An AI Public Service Announcement
1822
Why AutoGPT engineers ditched vector databases
AutoGPT decided to drop the support for vector databases: Pinecone, Milvus, Redis, Weaviate - all gone. What has happened to the original idea of efficiently storing agent's memories?
1823
LLMs can't self-correct in reasoning tasks, DeepMind study finds
A study by Google's DeepMind and the University of Illinois at Urbana-Champaign has found that self-correction in large language models (LLMs) isn't universally effective.
1824
LLMs confabulate not hallucinate
Minor terminological nitpick. People often describe the LLM as ‘hallucinating’ information whenever it makes up information which seems like it should fit for a given query even when it is trivially false. While evocative, this isn’t actually correct terminology. We already have a perfect word for this precise…
1825
Extracting Hacker News book recommendations with the ChatGPT API
2023-10-12: I have written a followup with some details about the process used to generate this list: Extracting HN Book Recommendations with the ChatGPT API, Vol. II
1826
Translating Latin demonology manuals with GPT-4 and Claude
At long last, a real-world use case for AI!
1827
Every database will become a vector database sooner or later
Why every database will become a vector database, sooner or later
1828
Weird A.I. Yankovic: a cursed deep dive into the world of voice cloning
What would it sound like if the musicians parodied by "Weird Al" Yankovic sang *his* parodies instead? I jumped off the deep end into the world of A.I. cover songs to find out, and the results are even worse than you imagine.
1829
Break-a-Scene: Extracting Multiple Concepts from a Single Image
A method for extracting multiple concepts from a single image.
1830
Our humble attempt at “how much data do you need to fine-tune”
Efficacy of OpenAI Fine-tuning API, Cost/Latency Considerations, Experiencing Catastrophic Forgetting, and...Getting Roasted by a Fine-tuned GPT-3.5