Oriol Corcoll

New work out now! 🚀 "The Hardness of Validating Observational Studies with Experimental Data" explores fundamental limits of combining experimental & observational data, and introduces a novel Gaussian Process approach. #AISTATS2025. #CausalInference #ML arxiv.org/abs/2503.14795

Spotify founder Daniel Ek explains why he doesn’t worry about competition from Google and Apple In this interview from 2015, Daniel Ek is asked how Spotify will compete against tech giants like Google and Apple who recently entered the music space. He responds: “Apple and Google do this and 10,000 other things. We’re specialized. We don’t do anything other than our own service. This is all I do every day. We think with that focus and the thousands - if not millions of hours - that we put into creating [our] experience, we will win.” He continues: “The way you win in this fast-moving world… is by being super focused on solving one problem better than anyone else and by moving faster than everyone else in solving that problem. It’s really simple if you think about it like that… Sooner or later you’re going to get defocused if you do a thousand things. You can’t do all 1,000 things super well.” Video Source: @twistartups @Jason

@TacoCohen Congrats!

Two weeks ago I joined Meta / FAIR, and I couldn't be more excited about this new chapter. Meta is indeed the only place left that supports highly ambitious long-term oriented & fundamental research projects and has a strong commitment to open science and open source. (and has hundreds of thousands of H100s lying around) In my short tenure here I've already met tons of super talented, ambitious and kind folks, and I'm blown away by all the futuristic tech being developed, as well as the sheer scale of Meta's operations. I will be joining @syhw & team to work on code generation. Codegen is obviously an important application domain, but it also provides a great setting for research on general methods for planning, search, reasoning, etc. Stay tuned!

Check out this thought-provoking piece by Léon Bottou and @bschoelkopf entitled "Borges and AI".

Thrilled to announce the first annual Reinforcement Learning Conference @RL_Conference, which will be held at UMass Amherst August 9-12! RLC is the first strongly peer-reviewed RL venue with proceedings, and our call for papers is now available: rl-conference.cc.

Is machine consciousness just around the corner? In our new paper, we approach this question from the neuroscientific perspective and explain why this is unlikely to be the case. just out in @TrendsNeuro w @mattlark & @jmacshine authors.elsevier.com/a/1hxp2botq09Yc

It’s #GAIA1 world, we just drive in it. Generate realistic driving videos using only prompts. See how it works below 🧵

Today is a big, big day for @Spotify. And for book lovers everywhere. Audiobooks are coming to Spotify Premium! 🎧📚

Do you dream of a world where some of the top podcasts would be spoken in your native language? Well, that’s now possible. We’re excited to pilot Voice Translation, a groundbreaking feature powered by AI that translates podcasts into additional languages—all in the podcaster’s voice. spoti.fi/AIVoiceTransla…

here is dalle 3, which imo is quite amazing: openai.com/dall-e-3 it will ramp to all chatgpt+ users over the next couple of weeks. fantastic work by @model_mechanic (head of dall-e), @neobjb @gabeeeegoooh @jingli911 (the other lead dalle ICs), and the entire team.

Language is the future for how we interact with robots. Today @wayve_ai is sharing a first look at LINGO-1, a new vision-language-action AI model. To give you a glimpse of its capabilities, here is a video of me playing with LINGO-1 yesterday morning.

@its_dibya @manan_tomar @vivek_myers Could you share the link to the paper?

Come check out our poster at 10 AM (that's now!) with @manan_tomar @vivek_myers in Preference-Based Learning (Room 316) about how RL agents can tap into large video datasets to explore efficiently in a new environment.

Steel drum collapse experiment. The water in the barrel is heated at its boilling point and sealed. The ice lowers the temperature and the internal pressure decreases due to condensation: the atmospheric pressure causes it to collapse suddenly twitter.com/Solocuriosos_1…

Counterfactual reasoning is vital for decision-making. Our @NatMachIntell paper on counterfactual inference & ML from @SpotifyResearch has been covered in this great @techreview article by @strwbilly. Check out this thread🧵 for an overview👇 surl.li/gfpqg

Another AI paradox: people are excited about LLMs, some even think that AGI is just around the corner. But some students are depressed how they can still get a PhD. Is it becoming pointless? Some personal notes on this. (1/8)

Stanford Alpaca, a new LLM player, has been fine-tuned with Nvidia A100 x 8 for 3 hours ($100) and augmented with OpenAI GPT-3 ($500). This model can be used on a single GPU or CPU, such as Apple Silicone or Raspberry PI, and offers performance similar to GPT-3 at a total cost of $600. crfm.stanford.edu/2023/03/13/alp…

RL matters for 3 reasons: autonom. data collection, non-diff. optimisation, and credit assignment over sequences. Now trivial: it's great for improving LLMs. Current RLHF collects data manually and benefits credit assignment. Going forward, data collection will be the key!

Introducing DJ ✨ An entirely new way to play your @Spotify, powered by AI. Now rolling out to Premium users in the U.S. and Canada. Head to Spotify now to meet your DJ.

JUST IN: Google invests $300 million in Anthropic as race to compete with ChatGPT heats up Anthropic was founded in 2021 by the team behind AI breakthroughs such as GPT-3 and Reinforcement Learning from Human Feedback (RLHF).

Our team at Spotify is looking for a research scientist intern (EMEA) to solve problems using causal inference. If you are interested or know anyone that might be, let me know! Official internship post here where you can find hard and soft requirements: lifeatspotify.com/jobs/summer-in…