How to organize an AI conference with Josef Lindman Hörnlund
The GAIA conference with Josef Lindman Hörnlund
Josef Lindman Hörnlund is a co-founder of Modulai and one of the organizers and co-founders of the yearly conference GAIA (Gothenburg Artificial Intelligence Alliance).
Josef, how did it all start?
After moving from Stockholm to Gothenburg back in 2014, after working in machine learning in Stockholm for a few years, I realized that in Gothenburg, very few companies were working with machine learning. So I joined a small meetup-group of people who discussed machine learning. We had meetups with the few companies operating within the field and ML students from Chalmers and Gothenburg University.
When we started, we were less than 20 people if I recall correctly. A few years after though, around 2016-17, it had grown a lot, and more than 100 people wanted to participate in each meetup.
We realised (or rather Daniel Langkilde, who was involved back then, did) that it made a lot of sense to organise a conference because of the interest. We had the first conference in 2018, and it was more or less just a big meetup. I even paid Lindholmen Science Park to rent the space out of my own pocket!
By 2019 we had put a proper non-profit association in place, were a little bit more mature, and we hosted the first “real” and more professional conference as the GAIA association.
What was your initial mission with the conference, and what are GAIA’s values now?
We wanted this event to be cheap so that students could afford to attend the event and to show what companies within Gothenburg are working with machine learning. We wanted it to be cool, inspiring, and community-building, and we are still holding onto this mission today.
How are you able to fulfill this mission?
As this small organisation, we need to cut costs and require financial and generous sponsorships from the larger enterprises in Gothenburg to run this event. At the same time, we don’t want this to be a conference where companies are selling their product or pitching a platform, like many other similar conferences. Every presentation is evaluated and assessed only on its content, so that the talks are interesting to practitioners and not sales pitches. It is a selling point and we need to hold on to it; otherwise, people won’t come, and we will lose our values.
What participants are you trying to reach?
We want this conference to have broad content and reach a wide audience within the machine learning community. We’re trying to cover most of the technical aspects of machine learning, some business and organisational themes, and of course data science and data engineering as they are essential parts for all companies working with machine learning in any way.
We want to have a variety of different topics throughout the day to offer at least one or two interesting talks to all attendees.
Do you have any other constraints?
One important challenge is getting female speakers. It is really hard, but we are working on being as diverse as possible. It’s a challenge that constantly comes back. It is tough in every technical field, but I think we can do a lot better than we are right now.
“I even paid Lindholmen Science Park to rent the space out of my own pocket!”
Josef Lindman Hörnlund
What are the future plans of GAIA?
Next year we’re celebrating our 5th conference, so we will try to do something special for that one. The machine learning community is rapidly getting more mature and experienced, and we need to adjust the conference to that. The number of companies in Gothenburg working with ML and with ML in production has been snowballing over the last couple of years, and it is clear that this also changes the requirements and purpose of conferences such as ours.
What are your biggest learnings?
The greatest challenge is to both listen to the audience and stay opinionated about how the conference should be. For example, this year, most talks were 20-25 minutes, and some think that is too short, and others find it perfect. It is hard to find a good balance here. Another challenge is to push speakers to be technical but not too technical.
What is the feeling when doing the event?
There’s a lot of stress before the event, both when selling tickets, getting speakers and making sure everything runs smoothly. The pressure is often released during the day and gives you a rush of energy which is really gratifying. It feels great to see all the happy attendees and get positive feedback afterward.
Who would be the ultimate speaker?
The point of this event is not to have the most incredible speaker on earth; it’s about the local scene and the competence we have within the city of Gothenburg. We want this to be a community-building network rather than a commercial conference. People can always watch their favorite speakers on Youtube, and it is not our job to compete with that.
Is there going to be a GAIA Stockholm edition?
I would love to see something similar in Stockholm, but the situation is quite different there with costs for conference halls a lot higher. I am not sure it would be possible to arrange something similar without going too “enterprisey.”
GAIA exists because of a fantastic team of people that spend a lot of their free time getting these events in place, with only the kick from organising a popular and hopefully important event as payment. A big shout out to Jakob Andersson, Amanda Nilsson, Niklas Antoncic, Johan Gustafsson, Daniel Sääf, Viktor Olsbo, Bissane and Nils Ingelhag who all helped organise this years conference.
Press here if you want to know more about GAIA.
Gothenburg Artificial Intelligence Alliance (GAIA)
GAIA is a non-profit association promoting the interest in artificial intelligence, machine learning, and data science in Gothenburg and the surrounding area. Founded in October 2018 by merging the organisation behind the Machine Learning and Data Science.
GAIA offers knowledge-sharing and networking through regular events. Meetups are held roughly bi-monthly and the conference is held annually in spring. Between events, members can engage through the GAIA Slack.