Pepper | Cutting Edge Computing

A Path to AI Explainability April 23, 2021

Posted by Peter Varhol in Machine Learning, Technology and Culture.
Tags: AI, explainable artificial intelligence, Pepper
1 comment so far

I mentioned a while ago that I once met Pepper, the Softbank robot that responds to tactile stimulation. This article notes that Pepper can now introspect, that is, work through his “thought” processes aloud as he performs his given activities. I find that fascinating, in that what Pepper is really doing explainable AI, which I have been writing about recently.

The result is that Pepper can not only carry out verbal instructions, but also describe what he is doing in the process. I don’t really understand how to code this, but I do appreciate the result.

Explainable AI is the ability of an AI system to “describe” how it arrived at a particular result, given the input data. It actually consists of three separate parts – transparency, interpretability, and explainability. Transparancy means that we need to be able to look into the algorithms to clearly discern how they are processing input data. Explainability means that we might want to support queries into our results, or to get detailed explanations into more specific phases of the processing.

Further, it appears that Pepper, through talking out his instructions (I really dislike using human pronouns here, but it’s convenient) is able to identify contradictions or inconsistencies that prevent him from completing the activity. That frees Pepper to ask for additional instructions.

That’s an innovative and cool example of explainability, and extends to the ability of the AI to ask questions if the data are ambiguous or incomplete. We need more applications like this.

I Want to Say Please January 1, 2021

Posted by Peter Varhol in Technology and Culture.
Tags: Alexa, Echo Dot, I Robot, Pepper, Robot and Frank
add a comment

Like many people, I have an Amazon Echo Dot. I don’t use it too much, and certainly not to its capability, but I’ve been using it of late to both check on packages and play Christmas music.

And like many people of a certain age, I want to say ‘please’ and ‘thank you’ to Alexa. Which raises very interesting questions (and I’m certainly not the first to ask them). I am a technologist who is rather adamant about not personalizing my tools. That’s simply what they are, of course.

On the other hand, on the surface level it appears that I’m having a conversation with Alexa, which strongly implies a certain level of civility. And in general I like to think of myself as a pretty polite person. So I say ‘Alexa, please . . .” and ‘Thank you.’

If it were simply me, this would be a very trivial question. It really doesn’t matter what I do in my interactions with Alexa. But I am concerned with more impressionable people think about their own interactions.

I had the opportunity to meet Pepper several years ago, the SoftBank robot. I confess to being a little standoffish with him (her?), but I do follow it on Twitter. My partner, on the other hand, glommed right onto him. It is programmed to pick up on tactile stimulation and facial expressions, and show a primitive level of emotion in response, and is quite impressive in that regard.

I was also watching the movie I Robot while on holiday. It didn’t get a lot of play, but it was quite well done (and I’ve been a lifetime Isaac Asimov fan; look him up).

I’ve also seen the movie Robot and Frank. Frank is an elderly thief whose son gets him a robot butler that he trains as a, well, a jewel thief partner. The robot becomes quite human in the process, and Frank is more of my generation, so I can appreciate his perspective. However, because I am a technologist, I am more wary about interactions with a robot.

I never grew up with any of this, of course. It’s ingrained in me to treat others respectfully when asking for something. I have heard that at least some children interact with Alexa as a machine rather than as a human, without courtesy, and I wonder who is right. To be fair, probably not me. But does such interaction spill over to dealing with humans? Perhaps that is why we seem like a somewhat less civilized society today than in the past.