Machine Learning Readings

Intelligence on Tap

I think this article does a very good job at outlining some issues with designing with machine learning.  My favorite part is when he addresses “AI coming of age.”  He really hits the nail on the head about the GPUs and massive data sets paving the way to allow for machine learning to really thrive.  However, I do think that this should raise some concern.  Consider who has massive data sets available to them.  I sure don’t.  The need for large amounts of data raises the barrier of entry for designers.  If we want to utilize AI in a project, we are limited by the data available to us.  This means that we can’t just add AI powered components to a product unless the data we have access to allows it.  A possible solution is to gather our own data, but this is a catch-22.  The device will need a large number of user to get the feature to work, but in order to get a large number of users, we need the working feature.  Therefore, I thoroughly agree that services to supply AI to the common man will be available in the near future, just look at Amazon Web Services and Google Cloud, they are already starting.  But I am still skeptical about how to integrate AI into everyday products due to the private nature of collected data.

UX Design Innovation

I think that the main takeaway from this paper is that machine learning is hard to integrate into design projects.  One thing I would specifically like to comment on is that it is difficult to prototype with AI.  Consider autocorrect on your phone.  This is a design feature that is deeply rooted in machine learning, and it has gotten better over they years.  But it didn’t start out that way.  It took years to train, and it doesn’t even work perfectly now.  I machine learning is to be implemented seamlessly, in a more critical position in some other project,  then the device simply would not work before the model model is trained.  This means that early in the design process the feature will seem less powerful than it really is because there is no way that a prototype has access to the same data as a product that has penetrated the market.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s