What comes after AI? Part 1.

Artificial intelligence is very hot right now, we have self-driving cars on the way, we have algorithms that detect faces, screen possible cancer treatments, translate languages, and even beat humans in Chess, Jeopardy and Go.

Elon Musk has started sounding the bell for more government regulation and warned of an AI-driven existential threat to mankind.

That seems like a lot, but is that the end? What could possibly come after that?

I will make the case over the next couple of posts that:

  1. Artificial Empathy (or Emotional Intelligence) is the next great leap for machine learning systems.
  2. That the building blocks for such a system are already in place.
  3. That the potential implications of such systems coming online are far greater than any prior technology revolution.

What do I mean by Artificial Empathy?

Well let’s start with the dictionary definition of empathy:

empathy: the ability to understand and share the feelings of another.

if we start there and add this video from Brené Brown on Empathy I think we get a good baseline to work from.

Now artificial empathy would be that ability to do this sort of connection at scale, and to a high degree of accuracy and utility.

How is that different from where we are now?

Where Artificial Intelligence can map stimulus to responses and then optimize, like

  • I show this ad, the user clicks the headline.
  • I show this in their feed, the user interacts with a retweet or a comment.
  • I send this email, this person or group will open it.

There is no measurement beyond a specific action was taken from the machines point of view. Providing analysis and picking between possible interactions is up to the humans. And while we may not understand the rules or predictive model that the AI system comes up with, it is still very transactional.

Doesn’t AE just mean sentiment analysis?

No, but that could be an input to it.

Sentiment analysis is not as deep as what I am imagining. If the processes described above are AI, and there is only quantitative data about a SIGNAL or interaction being generated, it does not care about the content of the signal. Sentiment analysis cares about the content of the SIGNAL. AE cares about the content of the SENDER. That is very different.

Sentiment analysis cares about the content of the SIGNAL. AE cares about the content of the SENDER. That is very different.

But I put it on a scale like this.

  1. AI –> Measure amount of responsive activities
  2. Sentiment Analysis –> Measure emotion of responsive activities
  3. AE –> Build up an accurate model of the internal emotional landscape of a person

Sentiment analysis tries to take that Quantitative Signal and convert it to some Qualitative Metric. So if there are 1000 comments on a tweet, how many based on word pairings and such are

  • Very positive,
  • Positive,
  • Neutral,
  • Negative, or
  • Very Negative.

This and some sort of summary of that data can be seen in the expanded, Like, Love, Angry, Wow, Sad reacts that Facebook now has.

Sentiment analysis studies the visible surface, AE works out the details of the inside of our minds and emotions.

So what does AE Look like?

This is a big deal, swing elections level big deal in its current state. Consider this coverage of the data firm behind Brexit and the Trump Campaign

In 2012, Kosinski proved that on the basis of an average of 68 Facebook “likes” by a user, it was possible to predict their skin color (with 95 percent accuracy), their sexual orientation (88 percent accuracy), and their affiliation to the Democratic or Republican party (85 percent). But it didn’t stop there. Intelligence, religious affiliation, as well as alcohol, cigarette and drug use, could all be determined. From the data it was even possible to deduce whether someone’s parents were divorced.

68 likes is an impressively small set of data, but it goes on:

Kosinski continued to work on the models incessantly: before long, he was able to evaluate a person better than the average work colleague, merely on the basis of ten Facebook “likes.” Seventy “likes” were enough to outdo what a person’s friends knew, 150 what their parents knew, and 300 “likes” what their partner knew. More “likes” could even surpass what a person thought they knew about themselves.

That is staggering to think about. And that was done when the only signal you could provide was “thumbs up”. With all the like/love/hate data, how much more accurate has that model become for working up your psychological profile?

I am sure you have a friend that if they ask you “how are you doing?” no matter how you answer they will know the real answer. Now imagine that level of insight into your feelings and emotions and actions is available to all your computer systems.

That is Artificial Empathy.

And that is what’s next.

In the next post, I will talk about all the ways that we are moving to make this level of Artificial Empathy and beyond possible.

2 Replies to “What comes after AI? Part 1.”

Leave a Reply