iiinova — where your digital self keeps evolving

I think the Pope should respond and quote the Bible, which has historical precedent even on Schmidhuber

020

As it happens, I believe both of them make good points.

074

These two men deeply believe what they say, but are also aware that everything they say will be interpreted through the viewers understanding of who they are and what they would be expected to say. They are, so to speak, symbols of themselves. Parodies of themselves, some would say.

381

Julian Togelius @togelius.bsky.social · 03/10/2026

This is the man who I did my postdoc with, and who taught me so much. No, not Leo, I don't know that guy.

2151

I was about to say the same thing, then I realized that I actually like both of them. Obviously, "like" does not mean that I agree with everything, but compared to the rest of the AI discourse...

170

Julian Togelius @togelius.bsky.social · 02/10/2026

Hello fellow human individuals, how are you doing?

0131

Reposted by Julian Togelius

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 30/09/2026

In our Nature paper, we introduce the first superhuman Stratego AI, which we built using general techniques that we developed for RL & test-time compute under imperfect information. www.nature.com/articles/s41...

nature.com

Scalable decision-making for games of imperfect information - Nature

Ataraxos, an AI for the board wargame Stratego, establishes a design pattern for reinforcement learning and search that is effective under large amounts of hidden information, a longstanding desiderat...

1225960

Julian Togelius @togelius.bsky.social · 30/09/2026

There's a new (free) book out by Media Evolution, and @ahmed-walid.bsky.social and I have a chapter in it. We're quite happy with it! www.mediaevolution.se/archive/A-Sh...

081

Reposted by Julian Togelius

Stephanie Chan @scychan.bsky.social · 21/09/2026

2/ We should be cautious about future scenarios involving systemic financial dependence. It's harder for financially dependent spouses to exit unsatisfactory marriages. And in countries where oil wealth funded subsidies and state employment, citizens have been vulnerable to state repression.

3101

Julian Togelius @togelius.bsky.social · 22/09/2026

Some limited personal evidence: for narrow enough domains, it’s entirely possible to finetune a small-ish model to be better than Astra and Fable. Can’t say more because it’s not an academic project

130

Julian Togelius @togelius.bsky.social · 18/09/2026

Might need an abliterated model

010

Julian Togelius @togelius.bsky.social · 18/09/2026

Seriously, has anyone made this work yet? Asking for a friend

110

Julian Togelius @togelius.bsky.social · 18/09/2026

Agentic AI could massively speed up productivity by doing all the mandatory compliance and harassment trainings for us

160

Reposted by Julian Togelius

Gautam Kamath @gautamkamath.com · 16/09/2026

Nihar Shah did a heroic experiment for TMLR: he spent 20-25 hours over two weeks interviewing authors of seemingly low-quality submissions about their own papers. He confirmed what we all suspected: people submitting these papers have *no idea* what is going on in them.

5293117

Julian Togelius @togelius.bsky.social · 16/09/2026

I had GPT review one of my papers in the style of H. P. Lovecraft. Did not disappoint..

030

It would also be the best move for safety, because literally everyone can study the models, find their weaknesses, and deploy them for defense. The only downside would be that Anthropic's and OpenAI's valuations would plummet. But I guess the rest of us could live with that.

1160

This naturally paces the frontier, because the economic payback on massive training runs mostly disappears if you have to open up your model to your competitors. It distributes power, and will lead to big advances in interpretability.

1140

Most effective AI regulation legislation would have terrible side effects, leading to loss of privacy, concentration of power, and restrictions on free research. But there is an alternative: regulate only closed-source (or closed-weight) AI. Keep open-source and open-weight AI unregulated.

24611

Somehow it feels relevant that Silicon Valley partly came out of the hippie culture, and that the hippie is one of the main nexuses of horseshoe-ism: the original hippie counterculture managed to be so radical and so conservative (even anti-modern) at the same time

030

Zach Weinersmith @zachweinersmith.bsky.social · 14/09/2026

Trying to visualize the multiple interlocking horseshoes of AI politics

140

Reposted by Julian Togelius

Meanwhile, the nearterm ai doom is real crew includes Elon Musk, Bernie Sanders, Josh Hawley, Sam Altman, and Amodei. And of course in my econ circles there's the "Let the glorious doom sweep away all economic enefficiencies" perspective.

131115

Reposted by Julian Togelius

Zach Weinersmith @zachweinersmith.bsky.social · 14/09/2026

Weirdest thing about ai today is that the group that believes that nearterm ai doom is bullshit includes: accelerationists, bluesky skeptics, Trump, the government of China.

1017115

Julian Togelius @togelius.bsky.social · 13/09/2026

Started my PhD in 2004, so I’m basically a trilobite

151

Julian Togelius @togelius.bsky.social · 13/09/2026

A few years ago, Georgios Yannakakis and I wrote an article about what to do as an AI researcher when a giant AI company steamrolls your research field. I just looked at it again and, well, it’s still relevant. ieeexplore.ieee.org/document/104...

0304

Julian Togelius @togelius.bsky.social · 12/09/2026

Brilliant. @aelkus.bsky.social can I steal this for lecture slides?

121

Julian Togelius @togelius.bsky.social · 09/09/2026

That does not seem strongly related. Interesting-looking paper, though.

100

Julian Togelius @togelius.bsky.social · 09/09/2026

New paper! We are honestly surprised that nobody seems to have explored this idea before. Results are interesting!

2102

Until the end of the year, I will continue to be a Professor at @nyuniversity.bsky.social. I will also continue in my role of Head of Research at Not1.

120

Here are the initial recruitment ads - please spread widely and encourage talented candidates to apply! Postdoc: www.chalmers.se/en/about-cha... PhD students: www.chalmers.se/en/about-cha... More rounds of recruitment to follow!

chalmers.se

Vacancies

196

Chalmers welcomes two new world-leading professors

I'm grateful to the Knut and Alice Wallenberg Foundation and WASP for the very generous grant that allows me to get a flying start on my new lab. Here's the announcement from Chalmers: www.chalmers.se/en/current/n...

chalmers.se

Chalmers is strengthening its research in materials chemistry and computer science through the recruitment of two internationally leading researchers. The initiative, worth a total of SEK 160 million,...

270

I am going to the country I was born in, Sweden. More specifically, my family and I will move to Gothenburg, where I will take up a professorship at Chalmers University of Technology and start a new world-class AI lab there. Yes, I'm hiring!

2171

Zach Gage @stfj.net · 07/09/2026

Big news! I'm leaving New York! ...but what on earth? Why? Where am I going?

3260

Reposted by Julian Togelius

Ok im gonna say something here that is going to draw a lot of different feelings from everyone, and I want to be clear i'm not "in support of this" or "against it". i'm just saying it as an observation that seems true to me: I think with the latest frontier LLMs, the “music” era of games is upon us

1014819