AI Hackathon in Norrköping

The AI boom has been a reality for a while now, and we’re probably somewhere around the ‘Peak of Inflated Expectations’ if we go by Gartner’s Hype Cycle.

There’s no doubt that AI is useful, and I don’t think many would argue that AI isn’t the future. But just how useful is it today, and what are its limitations?

This was the challenge our developers faced today.

Heading South

At lynes, most of us use some form of AI daily, especially our developers. To gauge the current state of AI and ensure everyone is up to speed, David, our CTO, planned an AI Hackathon in Norrköping.

As usual, I hopped into my fossil-free car and headed to Garvaregatan 4C. Upon arrival, I was greeted by the smell of freshly baked bread and coffee—no complaints so far!

We started with breakfast, during which David went over the day’s agenda, guidelines, and discussed AI assistants.

Before the event, he optimized several AI assistants in the code, generated some examples, and set up classic ‘no-go zones.’

The purpose of the day was to create a deeper understanding of AI and spark ideas for the future. David touched on topics such as:

• The importance of focusing on what AI can do here and now, not in a fictional future.
• How we can increase productivity for our end users by eliminating manual steps.
• How and when we should implement AI in the app.
• When we should NOT implement AI.

He left us with this thought:

AI, in this case, can be seen as a magical way to create a new feature from scratch in no time.

Time to Code

The developers formed teams, and luckily, I ended up with David. Some might think it’s cheating to compete with the judge, but since we only had one person with coding skills on the team, no fuzz were made.

To my surprise, there were no protests throughout the day, and I soon realized why. I was the only one viewing this as a competition!

Quite an unusual situation for someone competitive like me. Oh well, if no one wants to compete, we’ll win anyway. David and I used the classic pair programming method, which gave me the chance to check out the other teams, both to spy and to listen to the AI discussions.

As the afternoon approached, it was time for all teams to demonstrate their features.

‍

Successful Cases

As I mentioned earlier, the goal was not to develop groundbreaking features in a few hours. The aim was to gain a better understanding of how the technology could be implemented in our service.

Case 1

One team chose to implement AI to generate avatar images and background images for a user’s profile. It’s a simple and fun way to use artificial intelligence to create custom images directly in the app.

Case 2

Another successful case was the team that built a ‘chat summary’ feature. This allows you to jump into a chat, select messages, and get a summary of what you’ve missed in that channel. A very smart feature if you ask me!

Now, you might wonder why you don’t find any of these features in the app. Let me explain.

Before releasing a new feature, it goes through 1,000 different tests, security reviews, and more to ensure everything works as expected. This explains why none of the above features are available today.

‍

Take aways

There’s no doubt that the technology is mature and powerful. However, one team concluded that there are still blind spots, as today’s AI couldn’t replace the manual input needed for the function they chose to develop. When the necessary data isn’t available or the function requires too much manual input, it remains a manual task.

This was our first AI Hackathon. Keep an eye on the app, as we’ll be introducing even more AI-driven functionality soon!