Last day of UXLx and at this point I’m feeling overwhelmed because of the amount of information I absorbed in this 3 days. Probably will need some time to go through everything and set all the new information in my mind!
Today I decided to go to two completely different masterclasses, one about how UX and AI can work hand in hand, and another one about better research.
Apologies if these notes are getting shorter, but as the event is coming to an end, my energy is also lacking. Fridays am I right?
But getting to the point, this are my intakes on this day. Once again, please note that these are my intakes based on what I learned during the masterclasses.
This is a topic I was already acquainted to but it was refreshing to hear about it from Josh’s point of view. He presented great insights on how to use machine learning in our products.
Designing for what is next
In our everyday life we rely in machine learning based products several times. One of the most popular case is with Google Maps. How many times do you just follow the route that your GPS tells you and don’t even think about it? Maybe in a regular commute, where you know how to go from A to B but you still rely on the maps to navigate around traffic?
How many times did it send you to a shady route saying that it’s faster and you just went with it? I bet some.
This provides the following question:
Do we want to have machine learning defining how/ where traffic builds?
Do we rather have machine learning helping us making better decisions without deciding by us what to do?
How do we use machine learning in a way that amplifies human decisions
Machine learning in small tasks
There are different types of ML we can find in apps we use on a daily basis. Let’s take a look at them
- Recommendation: Slack that matches subjects to inputs in search
- Prediction: Keyboard that predicts words you may pair after based on what you write
- Classification: Google form that suggests an answer type based on your question
- Clustering: Sort of the same of classification but only done by ML (has no human intervention)
- Generation: Describing a web component and the system develop it and making it functional (text to code)
Machines are weird… Because humans are weird!
Machine Learning potential issues
ML can be a helpful tool to have in mind in our close future, but it needs to be tended by the humans.
Unfortunately bias is everywhere in human behaviour, but it can be a good thing when they surface because that allows us to act on them (of course this would be best if they didn’t exist in the first place).
Machines can amplify our work, augment our perception
Let’s use Machine Learning as a Design Material