Dr. Michael Green retweeted
We are just days away from officially launching Outpost and we are taking in applications for our first company drop!
Dr. Michael Green retweeted
Working on gamifying and monetizing LinkedIn InMails—finding a way to take high skilled workers and give them passive income. "Gig Economy" but for white-collar workers
Dr. Michael Green retweeted
Replying to @Jason
Dr. Michael Green retweeted
I'm glad that everyone has now read Ted Chiang's essay and so has caught up on the AI art debate in the Computational Creativity community from a decade ago.
Early in a technological adoption! I’m very curious and excited to see “what are the good use cases” for LLMs. I’m not sure we’ve truly cracked it with just autocompletion just yet. Surely there is more to unlock here :) just takes time
This essay by Ted Chiang is really good and makes many points I agree with.
newyorker.com/culture/the-we…
Note that it is possible to say this while also thinking that LLMs and diffusion models are amazing technical breakthroughs that will (eventually) revolutionize society. It's about seeing the technology for what it is, and finding the right use cases. Right now, it feels like we're at a stage similar to when people tried to sell personal computers in the early 1980s with the argument that they could help you organize your recipe collection.
Dr. Michael Green retweeted
Democrats should introduce a gun control bill called the "The Donald Trump safety and gun control bill" and make republicans vote against it.
Dr. Michael Green retweeted
We (@Amidos2006, @Bumblebor, @Smearle_RH, @err_more, @togelius) wrote a chapter about evolutionary machine learning and games as a part of the handbook on evolutionary machine learning. I hope you enjoy it.
link.springer.com/book/10.10…
I love this!
What if we set GPT-4 free in Minecraft? ⛏️
I’m excited to announce Voyager, the first lifelong learning agent that plays Minecraft purely in-context. Voyager continuously improves itself by writing, refining, committing, and retrieving *code* from a skill library.
GPT-4 unlocks a new paradigm: “training” is code execution rather than gradient descent. “Trained model” is a codebase of skills that Voyager iteratively composes, rather than matrices of floats. We are pushing no-gradient architecture to its limit.
Voyager rapidly becomes a seasoned explorer. In Minecraft, it obtains 3.3× more unique items, travels 2.3× longer distances, and unlocks key tech tree milestones up to 15.3× faster than prior methods.
We open-source everything. Let generalist agents emerge in Minecraft! Welcome you all to try today: voyager.minedojo.org/
Paper: arxiv.org/abs/2305.16291
Code: github.com/MineDojo/Voyager
Deep dive with me: 🧵
Dr. Michael Green retweeted
Dear Students, please don't get discouraged... we need you. Quoting another Geoff: "The future depends on some graduate student who is deeply suspicious of everything I have said" (8/8)
Dr. Michael Green retweeted
Our game with @d_nielv , @RgKrista , @Strouthas , @emmanuela_88 , Yiannis Brellas Naomi Cutajar and L., "Valletta: Streets of History" was featured on @TheTimesofMalta today! 💙
Still can't believe that we are only 3 days away from the launching event! 🙀
Dr. Michael Green retweeted
Ok, enough Sokoban. I know you care more about Super Mario. Therefore, on to the next paper, by our good friends at @RoboEvoArtLab at @ITUkbh
Dr. Michael Green retweeted
Large language levels can do... lots of things. But can they generate game levels? Playable game levels, where puzzles are solvable? Two papers announced today address this question. The first is by our team at @NYUGameLab - read on for more:
my first 5 tweets in my timeline is El*n.... wtf is this?
All from out-of-order times too: 33m, 2hr, 45m, 1hr...


























