Mankind can create machines to do work. How could we define passion in artificial intelligence? How could we define a passionate AI?

Would passion cause the machine to do a better job? How could we compare the performance of a passionate machine, as opposed to a non-passionate one? How could we measure the level of passion?

  • $\begingroup$ You're asking too many questions in this post. $\endgroup$
    – nbro
    Commented Nov 17, 2019 at 0:18

2 Answers 2


An elementary approach to 'passion' would be to pre-assign different areas for the program to be 'passionate' about and associate different numeric 'drive strengths' with each (perhaps adaptively). Mechanisms of this sort were studied in Toby Tyrell's widely cited PhD thesis on 'Action Selection in Animals'

More recently, some more sophisticated AI architectures have been developed under the heading of 'Intrinsic Motivation'.

Here is a link to a paper on the subject by Pierre-Yves Oudeyer, a leading expert in the field of Developmental Robotics.

With regard to the question "would this cause the machine to do a better job?", that would very much depend on how open-ended the architecture is:

It's clearly easier if, rather than having to spell everything out in detail to a machine, we can simply specify a problem at a high-level and let its own motivations cause it to explore promising avenues.

Conversely, if motivations are too open ended, it may well spend all its time doing the equivalent of 'doodling on its paper' (Hofstadter).

Hence, like people, the quality of the output will be a function of its internal dispositions and could be measured in the same way for a given task (e.g. quantitatively for scientific activities, qualatatively for the arts).


Interesting question.

Well if you really think about it, what is passion? How does that passion comes to be a passion. One of the main topics you might want to touch here is conditioning and thus motivation.

Think about the following:

I have a passion for programming

Why do I have a passion for programming?

Because when I wrote my first program I was positively reinforced by the fact that I completed a program, I was negatively reinforced because I removed my frustration of not completing the program

How come that I have gone through that programming frustration and stick to it even if I was frustrated?

Because I wanted to learn programming

Why did I wanted to learn programming?

Because I wanted a light on an arduino to turn on (projected reinforcer)

Why did I wanted to turn the arduino light on?

So I could learn programming and because I though it was cool (classical conditioning association that will later be reinforced, projected reinforcement happened right after the classical conditioning association between turn on a led happened)

This can be done through a neural network, where each association is reinforced through a probability of outcome For example, I did learn arduino, on purpose because it seemed the easiest way to start coding, so the probability of positive outcome was high

This about an opposite situation Let's say I do not know calculus, and I barely know elementary algebra, if someone started to teach me about integrals saying that this is the only way to start learning more math, I will not be motivate to do so because since I cannot even conceptualize what an integral can be, it will be really hard for me to understand it thus I will not learn calc

Thus we can also discern that motivation is reinforced in small behaviors

Another more practical and realistic example you might use is

If you trow a rat in a cage, and make him lever-press do you think he is going to? No. Although if you reinforce the behavior of going next to the lever slowly and at the end he will lever press and you then reinforce that behavior he will.

Thus, passion is compartmentalized, and that's what you have to do in your NT and make it mathematically


Small hint, it's a progressive function


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .