Mikael Hansson

I’m Dr. Bonnie Barstow but my Peppers are not KITT

When I was a little girl I tried to watch every episode of the American television series Knight Rider, not because of the male main character who fought injustice and crime, but because of Dr. Bonnie Barstow, the female chief technician of KITT. KITT was an artificially intelligent self-driving car that could understand and communicate in natural language, and one could remote-connect to it via a wrist watch.

Which 10 year old’s heart wouldn’t start beating faster by this? I wanted to be like Dr. Bonnie Barstow when I grew up, building technology and machines that were so intelligent, that humans could talk to them!

Eventually, I started my university education and focused on natural language processing or how to make machines understand and communicate in natural language. Back then (about 20 years ago) we studied linguistic theories in order to understand the underlying principles of how humans process natural language, and then developed so-called finite state methods (e.g. formal grammars, graphs and automata) that could be implemented in machines.

Since recently, everyone in natural language processing (or in any other research area of your choice) seem to use machine learning methods to solve problems: given a problem statement x and a machine learning algorithm y, apply y to x, and obtain very good and fast results. The only tiny remaining question is: why do we obtain these results?

Do machines today speak and reason like humans do due to the omnipresent machine learning methodology? No, they don’t. Did we make progress? Yes and no. We obtain faster and better results for specific problems, but those who want to understand some principles of human cognition, do not gain the scientific insight they seek.

Currently I’m working with colleagues at the Department of Computing Science, and three Pepper robots on developing dialogue management approaches that combine machine learning methods and finite-state methods. Machine learning algorithms serve us well for problems that involve mining patterns or correlations over large data sets, and finite-state methods give us the understanding that we, as scientists, naturally seek. I believe that the key to a new advancement in AI is combining methodological approaches, so that we get reliable and fast results, combined with insight into stated problems and their solutions.

I want to share an advice by Marvin Minsky (one of the pioneers of AI) who said ”If you just have a single problem to solve, then fine, go ahead and use a neural network. But if you want to do science and understand how to choose architectures, or how to go to a new problem, you have to understand what different architectures can and cannot do”.

Over the past years, I obtained my PhD, tried out various methodological approaches, worked with Pepper robots, and I can say, that I am Dr. Bonnie Barstow but my Peppers are not KITT. Not yet.

Photo: Mikael Hansson

 

15 Kommentarer
  1. Johanna Björklund says:

    What a nice blog! Keep it going, and soon we’ll be in an age when the following quote from Knight Rider doesn’t seem wholly like SciFi:

    Maddock: The Knight 4000 is heading into first time combat, and you can’t tell me if it’s functioning properly?!

    KITT: Well no, unfortunately we may have to rely on HUMAN JUDGMENT!

    Svara
  2. Kai-Florian Richter says:

    Nice blog! But your Peppers need many more blinking lights 🙂

    More seriously, I would agree that especially for such (almost) human-like artificial agents it is crucial that their behavior is predictable for layman users and understandable to ‘the experts’

    Svara
  3. Carl-Erik Engqvist says:

    A really good text from the ”inside” of the development of AI! I especially liked the statement: ”I believe that the key to a new advancement in AI is combining methodological approaches, so that we get reliable and fast results, combined with insight into stated problems and their solutions” and its connection the Marvin Minsky.

    Maybe a long shot, but the statement reminds me of when I was a student at Lunds university. There was a lecturer in the science of law that encouraged and even enforced his students to read a huge selected list of both classical
    and modern novels. When questioned by his students, why a law student
    should read fiction instead of just law texts, he replayed as they will be
    working with people and problems related to the desires of people, the
    best source (both cheaper than travel and more environmentally friendly) for gaining multiple perspectives and a deep insight into the human condition is
    reading loads of good fiction (and look at art).

    Keep up the good work! Looking forward to more posts 😀

    Svara
  4. Anna Fola says:

    I love your drawing and I’ve always wondered- is Michael Knight going to lend me his car when I am older?- 🙂 Many many inspiring questions- the future is here!

    Svara
  5. Kalle Prorok says:

    Nice blog. I’m sorry, can’t relate to Knight Rider and Barstow but we had Star Trek, Moonbase Alpha and others;). One joke within Neural-Net people groups is that language projects involving Linguists performs worse than without, I guess it is certainly not true, particularly if you want to understand what’s happening. I play around with Reinforcement Learning and that is really fascinating; maybe you should have a look at it if you haven’t already. Good luck!

    Svara
    • Suna Bensch says:

      Kalle, thanks for your comment! A similar joke, that is popular among linguists, is by Frederick Jelinek ”Every time I fire a linguist, the performance of the speech recognizer goes up” 🙂 Good luck to you too!

      Svara
  6. Lennart Johansson says:

    Hi, agree it is a nice blogg though I never really identified myself with any of the caracters in The Knight Rider. KITT was just a tool, though an advanced tool, but it did not have its own agenda and is that not the thing that scares people about the AI development? Instead of KITT we get Christine?
    And speaking about the connection to SF – we have a lot of concepts deriving from Isaac Asimov and his ”Laws of Robotics” that has had a huge impact in both litterature and the film industry. Several scenarios which he develops, concerns the differences between actions driven by ”cold logic” vs emotions and will it become necessary for a true AI to develop emotions in order to understand us humans?
    Lennart

    Svara
    • Suna Bensch says:

      Lennart, thank you for your comment! Interesting thoughts about SciFi and emotions. SciFi can even be an inspiration for AI researchers helping us to dream beyond the limitations of our current algorithms but also warn us about ethical consequences. Terminating now, but I’ll be back 🙂

      Svara

Lämna en kommentar

Kommentera

E-postadressen publiceras inte. Obligatoriska fält är märkta *