From Punching Paper to AI Enablers

Computers have evolved in mind-blowing speed. Not too long ago, programmers punched holes in cards, now they are using AI enablers to get the job done. A look on how far we have come.

Computers have evolved in mind-blowing speed. Not too long ago, programmers punched holes in cards, now they are using AI enablers to get the job done. A look on how far we have come.

My mother was a programmer. I remember playing with the used paper that was fed into the mainframes. Rolls of tens meters long that looked like Emmentaler cheese. I later learned that those holes conveyed information into the heart of the machine. To say that was a cumbersome job would be an understatement.

Things evolved a lot since then. When I started programming myself, I only had a keyboard. How thrilled I was when using my first mouse! To select what I was interested in without writing green text on a black screen – it was amazing. Then came bigger screens followed by the interfaces themselves, all making interaction easier still. The next revolution did not take long: Today, I am able to talk to my phone and it answers my questions and helps with daily chores. Now, I am wondering: What is next? AI-based interfaces are my guess.

Unspoiled user experience

AI is changing human-computer interaction yet again. Companies like Swisscom are working on making the user’s experience seamless. The best interface is one that is barely perceived. Thanks to AI enablers, transmitting a message will take close to no effort. The user will simply be talking to the computer and it does the rest. It’s no longer an idle wish: We can now deliver systems that understand spoken language.

The AI scientific community has made major breakthroughs in recent years. In the area of speech processing these breakthroughs are most obvious. Today, computers are able to understand spoken words better than another human would. However, this is only true for English, some accents not included, and not in all domains.

These constraints become research questions that Swisscom, together with its academic partners EPFL and Idiap are working on.

Four reasons why developing AI enablers is great news:

  1. Let’s say you speak Italian. However, most of the previous relevant cases were recorded in German. Wouldn’t you want to profit from that knowledge and have the solution translated to you? You bet you would. To deliver these kind of machine translation systems, we are working with the Natural Language Laboratory at Idiap.
  2. Swisscom, together with Idiap, are working on delivering speech recognition for Swiss languages at a level that allows spoken interaction.
  3. Let’s pretend you have a problem with your mail client. Until now, you were probably googling for a solution. Think of asking Swisscom what to do instead. The system will scan cases similar to yours and provide a possible solution before you can say “coffee”. The answer will not be delivered by a person but by a Chabot. We are currently working with the AI lab at the EPFL to deliver a bot with these capabilities.
  4. Further down the road, we believe that all interaction should be personalized. Personalization should include all the data that YOU choose to share about yourself. Bills, installed software, your favorite color, how your dream car revs – the more data we are able to include in the Chabot’s brain, the better it tailors its answers to your specific needs.

To sum it up: Interaction should no longer be perceived – it should come naturally instead. Searching should no longer be perceived – all the relevant data should be there, ready to be retrieved. Personalization should not be a feature – it is a given. That is our promise for the future. And this excites me just as much as my mom’s Emmentaler cheese paper did back then.