Messages
1/3/2023
11 : 54
Import AI 313: Smarter robots via foundation models; Stanford trains a small best-in-class medical LM; Baidu builds a multilingual coding dataset
If AI timelines keep shortening among those working directly on AI systems, at what point do you need to 'hard update' policymakers that something strategic and important is closer than it
12/12/2022
12 : 24
Import AI 312: Amazon makes money via reinforcement learning; a 3-track Chinese AI competition; and how AI leads to fully personalized media
2022 will probably be the 'eternal september' of AI research; everyone is starting to pay attention View this email in your browser Welcome to Import AI, a newsletter about artificial
12/5/2022
11 : 34
Import AI 311: Distributed GPT busts the political economy of AI; Apple optimizes Stable Diffusion; AI war startup raises $1.48 billion
We live in an age of wonders, able to conjur up engines of creation that weave their own idiosyncratic synthesese from the threads of civilization. The West, after its dabble with a secular culture, is
11/28/2022
12 : 24
Import AI 310: AlphaZero learned Chess like humans learn Chess; capability emergence in language models; demoscene AI.
If I wasn't spending my dayjob working directly on AI and AI policy, I'd be running a scaled-up version of Import AI. Lately, I've been wondering if there's a way to do both - it'd
11/14/2022
14 : 4
Import AI 309: Generative bias; BLOOM isn't great; how China and Russia use AI
If we wanted to make the next five years of AI development go well, what would be the three most important things to work on, and what should be deprioritized? View this email in your browser Welcome
10/31/2022
10 : 4
Import AI 308: Recursively self-improving LMs (!!!), 3.1TB of code data; DALL-E2 makes alien errors
Honestly, these days I feel pretty confused about AI. AI progress is happening at such an astounding rate I find myself asking 'why isn't basically everyone working on this?'. I don't
10/25/2022
15 : 4
Import AI 307: Copilot lawsuit; Stability raises $101m; US v China CHIPLOMACY
If all AI research stopped today (but engineering and improvement of existing systems continued), then how would the world look in a decade? View this email in your browser Welcome to Import AI, a
10/17/2022
10 : 34
Import AI 306: Language models learn about the world via MuJoCo; Amazon releases a big Q&A dataset; and DeepMind tests out multimodal systems
In the same way dogs and whales are alien intelligences with respect to humans, how 'alien' might AI seem to us? View this email in your browser Welcome to Import AI, a newsletter about
10/11/2022
11 : 24
Import AI 305: GPT3 can simulate real people; AI discovers better matrix multiplication; Microsoft worries about next-gen deepfakes
If I can be simulated by GPT3, then will a larger LM simulate me better than myself? View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence. Forward this email
10/3/2022
14 : 14
Import AI 304: Reality collapse thanks to Facebook; open source speech rec; AI culture wars.
Are the culture wars we're experiencing in the AI industry similar to those which occurred at the beginning of the rail and oil industries, or are these culture wars radically different? View this
9/19/2022
14 : 4
Import AI 303: Adversarial examples for language models; Censorship vs 'Safety'; free image classification from the StableDiffusion people
How many empires will be built with AI as the fundamental input resource? View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence. Forward this email to give
9/12/2022
14 : 14
Import AI 302: Fictional AI labs and AI theft; Google makes an audio model by training like a language model.
What predictions about future AI progress are most likely to be wrong? View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence. Forward this email to give your
9/6/2022
13 : 14
Import AI 301: StableDiffusion; CHIPXODUS; Microsoft makes a big bet on pre-training
What kind of values might the Egyptians and Romans and Aztecs encoded into their AI systems, if they had the chance to build them? View this email in your browser Welcome to Import AI, a newsletter
8/22/2022
13 : 4
Import AI 300: Google's Bitter Lesson; DOOM AGI; DALL-E's open source competition StableDiffusion
Once AI systems can retrieve from external knowledge bases, they will become true cyberlibrarians View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence.
8/8/2022
9 : 54
Import AI 299: The world's best language model is Made in China; NVIDIA boosts LLM training; OpenAI shows how to 'fill in the middle' on a LM
What will be the first new thing a superintelligence will invent? View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence. Forward this email to give your chums
7/25/2022
13 : 24
Import AI 298: Mimetic models; LLM search engine raises $25m; UK splits from Europe on AI regulation
How many parameters will be tunable in the 'default settings' of a superintelligence? View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence. Forward
7/18/2022
15 : 4
Import AI 297: Ukrainians add object detection to killer drones; YOLOv7; and a $71,000 AI audit competition
In the same way humans have a notion of the aesthetics of abstract sciences, like theoretical physics or advanced mathematics, might computers develop notions of 'beauty' at even higher levels
7/11/2022
15 : 4
Import AI 296: $100k to find flaws in LLMs, NVIDIA uses RL to make better chip parts; + 256gb of law data, and a story about the cyber gerontocracy!
Will we ever have 'old growth' computers in the way we have 'old growth' forests today. In the same way there are mainframes that have been running (albeit with parts swapped out and
5/20/2022
15 : 4
Import AI 295: DeepMind's baby general agent; NVIDIA simulates a robot factory; AI wars.
If it is possible to develop human-level AI, at one point will we make the first AI magician that perplexes even the most accomplished human magician? View this email in your browser Welcome to Import
5/10/2022
13 : 14
Import AI 294: China makes a vast facial recognition dataset; Facebook releases a 30bn parameter model; real world RL
If an AI designed the world we lived within, would we know? View this email in your browser Welcome to Import AI, a newsletter about artificial intelligence. Forward this email to give your chums an AI