Past Events

Some of these events where organised before AI Safety Support existed as a org/brand, but the ideals behind AI Safety Support is older than the name. The same motivation that created AI Safety Support also created all of these events, that's why they are on this list.

AI Safety Discussion Days
Regular events | September, 2020 - Present
Zoom and Icebreaker

Participant feedback:
"Interesting talk! And it was great chance to see other AI safety researchers."

"Learnt about Xuan's really cool research (which is relevant to my PhD project), met some cool people, and got sent some further relevant research that hasn't yet been published"

"Several fruitful discussions!"

Online LessWrong Community Weekend
September 11-13, 2020
Discord, TeamUp and more


This was not an AI Safety event and does therefore strictly speaking not count as an AI Safety Support project. But I still want to put it here becasue it's my (Linda's) biggest online event to date, and running it was a useful experience, and also an event I'm proud of.

Participant feedback:
"I am so glad this happened. It was a pleasure getting to meet people with interested ideas and knowledge from all over the world."

"A far more supportive and less confrontational community than I'm used to. There were many minor positives, but that was the single biggest thing."

"Discord as a main platform is nice"

AI Safety Discussion Days
June 7, July 18 and August 17, 2020
QiqoChat

Participant feedback:
"Made some notes for further thinking, learned about new stuff, got references for additional reading, understood some things better, shared some ideas or knowledge."

"Some face time to discuss with friends and keen people about things I am interested in"

"Some clarifying discussion on ML-based alignment techniques and factored cognition"

Web Technical AI Safety Unconference (Web-TAISU)
May 13-17, 2020
QiqoChat

Participant feedback:
“I got a better sense of the gaps in my knowledge about AI Safety; I became more motivated to engage with AIS (at least temporarily, no idea how long-term this will be); I met some people I might stay in contact or collaborate with (though the value of that will strongly depend on whether that actually happens)”

“Feedback on writing style blog posts (made it clear that I want to transition quicker to making clear distinctions in next posts) and Daniel Demsky's offer to comment on drafts. Also picking up what's happening in terms of insights + people active in the community.”

Learning-by-doing AI Safety Research workshop (LAISR)
October 24-27, 2019
EA Hotel, Blackpool, UK

Participant Feedback:
“So my experience at the workshop was: getting some very high quality teaching the first day, and then thinking about problems through. After trying to sketch my own solutions, the ideas and research agendas which we learnt about afterwards made more sense.”

“Meeting and working with great like-minded people and learning so much from them. Before the workshop, I was already into EA and AI safety, but this workshop gave me the feeling of being really introduced to the community and inspired me even further. It has definitely impacted me in a positive way and I feel like I will look back to it in the future and consider it a defining moment in my early career.”

Techical AI Safety Uncoference (TAISU)
August 22-25, 2019
EA Hotel, Blackpool, UK

Participant feedback:
“Provoked me to actually develop my ideas, heard about 20 different leads, found 2 possible shovel ready projects, met several people I need to know, had lots of fun.”

“Best value for money/effort of any conference I've attended yet”

I got a lot of great ideas out of it. similar in value to me to attending EAG. similarly hard to other ways i would achieve this via attending conferences. there's not really a replacement that exists at the moment for something like this given that i don't work at an org where i get to have in person conversations about AI Safety research frequently.”

Learning-by-doing AI Safety Research workshop (LAISR)
August 16-19, 2019
EA Hotel, Blackpool, UK

Participant Feedback:
"A good big-picture view of the problems in AI safety. Practice in formulating ideas and thinking about these concepts. Interesting discussions outside the main workshop."

I learned some new ML material, including new interpretability work. I was exposed to a lot of different views on how to tackle AI safety, including at least two new routes that seem promising to me. I think I've perhaps become a bit more open-minded about different possible paths. I also met a lot of interesting people that I might be able to collaborate with in the future.

AI Safety Camp (In collaboration with others)
April 12-22, 2018
Las Palmas, Gran Canaria


This was the first ever AI Safety Camp.

Research output