How Procrastination Can Be Useful For AI Systems, Including Self-Driving Cars

Dr. Lance Eliot, AI Insider

Image for post
Image for post

As the age-old joke goes, when someone asks you what the word procrastination means, you are supposed to say “I’ll get back to you about that.”

I’m sure we’ve all felt like a procrastinator at one time or another. Often considered a negative aspect of human behavior, some liken procrastination with being lazy, careless, and otherwise less desirable than being prompt and proactive.

We might though be somewhat making a false generalization about procrastination.

Does being a procrastinator always have to be bad?

It is said that prolific and ingenious inventor and artist Leonardo da Vinci was known for dragging out the works he owed his patrons and often taking nearly forever to get done what he had been obligated to produce. Charles Darwin had acknowledged that he often put off things that he was supposed to do or wanted to do. Few realize that he took years to write his acclaimed “On the Origin of Species” and was at times using his time to instead study barnacles. If these greats had bouts with procrastination, can it really be that bad a thing?

Psychologically, it is theorized that people often procrastinate because they fear the act of doing something that might fail. As such, in order to avoid failure, they postpone it.

For some people, they are an occasional procrastinator. Perhaps most of the time they get things done on a timely basis. A particular situation might cause them to go into a procrastination mode.

Sometimes a procrastination can have potentially dangerous consequences. Not taking care of the failing brakes on your car, due to procrastination of taking your car to the auto repair shop, could have deadly results.

Theories About Our Procrastination

There are the perennial or serial procrastinators. These are the types that just seem to shove everything off into the future. No reason to get something done today, if you can hand it off to the future, they believe.

There’s a well-known theory that somewhat covers this, called Temporal Motivation Theory (TMT). You might find of interest a core formula often used to express TMT:

Motivation = (Expectancy x Value) / (1 + Impulsiveness x Delay)

Your “Motivation” is the amount of desire that you have to achieve a particular outcome. If your motivation score is low, you are more likely to procrastinate. If your motivation score is high, such as if you realize that your brake light being out is putting you in grave danger, you are more likely to take action about it. We can calculate your motivation for a given circumstance.

The “Expectancy” is the probability of achieving success on the matter at hand. The “Value” is the reward that you personally will gain by achieving the desired outcome. By multiplying the Expectancy by the Value, the formula is saying that if one of those variable is low it is going to bring down the combination of them, while if they are both high it will make their combination higher. I think that my expectancy of fixing my brake light is quite high (just need to get the car over to the repair shop), and the value of increasing my safety while driving my car is high.

The “Impulsiveness” is the person’s sensitivity to delay. Some people are very impulsive and need to do things right away. Other people are more prone to taking their time or at least considering that they are willing to take time and don’t need to immediately handle the matter. There’s the “Delay” which is considered the time to realize the needed achievement.

Anyway, it’s kind of an interesting formula because it tries to mathematically express something that we all seem to know is happening, but don’t have at hand a tangible way to calculate why we do what we do.

People often make excuses for why they procrastinate. When I refer to these aspects as excuses, I should clarify that maybe they are valid. We often react to the word “excuse” and think it is a made-up aspect or an attempt to deflect blame.

Here’s some of the traditional excuses or coping responses:

  • Avoidance of the matter
  • Denial about the matter
  • Trivialization of the matter
  • Distraction about the matter
  • Mocking of the matter
  • Blaming about the matter

Procrastination Can Be A Manifest Strategy

This quick introduction to the topic of procrastination then brings us to a major final point before I move onto using this foundation for other purposes herein.

I claim that procrastination can occur perchance, but it can also be a manifest strategy.

I’ll highlight this point by next discussing how procrastination within an AI system can be at times beneficial, while at other times be less so.

Applying Procrastination As Part Of An AI System

What does all of this have to do with AI self-driving driverless autonomous cars?

At the Cybernetic AI Self-Driving Car Institute, we are using the core aspects of “procrastination” for two purposes, one is serving as a direct strategy of the AI driving the car, and the other is to deal with what we believe will be a human foible regarding AI self-driving cars.

Let’s tackle the human foible topic first.

As I’ve said many times, an AI self-driving car is still a car. By this I mean that some people are getting into their heads that an AI self-driving car will magically work 24×7 and will never have any mechanical problems or breakdowns. A car is a car. The brake lights are going to go out, just like on a regular car. The oil will need to be changed. The transmission will need to get overhauled.

This is going to actually happen more frequently and with deeper impact since we are expecting AI self-driving cars to be running all the time, versus today the average car is unused nearly most of the day.

Not only will the conventional parts of the self-driving car breakdown, but you can bet that the specialized add-on parts are going to breakdown too, such as the added sensors, processors, etc.

Now imagine an AI self-driving car that Joe Smith has purchased and he’s using it for himself, for his family, for his friends, and renting it out too for ride sharing. That self-driving car is busy.

When something snaps or breaks on the AI self-driving car, we need to consider these ramifications:

  • Does the AI self-driving car realize that something is broken or amiss?
  • Can the AI self-driving car continue safely operating?
  • Is there anything the AI self-driving car can do to get itself repaired?

We cannot assume that the AI will even know that something on the self-driving car is broken.

Example In The Use Case Of Autonomous Cars

Here’s how procrastination comes to play.

Suppose the human owner becomes aware that some aspect of their AI self-driving car has gone afoul. Maybe an occupant riding in it called the owner to complain. Maybe the AI detected that something was amiss and texted the owner to indicate that the self-driving car is having a problem. What will the human owner do?

You might assume that the human owner will promptly make sure that the AI self-driving car gets repaired. Perhaps so, perhaps not.

Will the next occupant that gets into that AI self-driving car even know that it is only operating with nine cameras? Possibly not. Should we ensure that the AI of the self-driving car warns any passengers about any anomalies? The owner of the self-driving car might not like that idea, and be worried that as a ride sharing rental that they’ll lose money if the AI starts to blab about what is wrong with the self-driving car.

There aren’t any regulations that force the AI to reveal what’s going on. The owner likewise is not under any direct law to do so, though you could construe various aspects about safety and the public that might turn this into a crime. For the moment, the use of AI for self-driving cars is so new that we have yet to figure out what twists and turns will happen, and nor do we yet know what kinds of regulations and new laws might be required.

The overall point is that we need to anticipate the dangers of potential “procrastination” based on human foibles and how it might adversely impact AI self-driving cars and their safety on our roadways.

Leveraging Procrastination In Driving

The second part of the procrastination aspect related to AI self-driving cars involves using procrastination as a AI driving strategy.

I know that you might be somewhat surprised or shocked at this idea of using procrastination purposefully. Remember though that I earlier stated that procrastination can occur by happenstance, or it can be used as a directed strategy.

Suppose I’m driving on the freeway. My exit is up ahead. I do the right thing and long before the exit make my way over to the rightmost lane. I sit there in the slow lane, maybe a mile ahead of my exit. I’m going bumper-to-bumper but it’s Okay because at least I know I am securely in my needed exit lane. Do humans really drive this way? Some do, many do not. What’s actually more likely is that I’ll “procrastinate” and put off getting into the slow lane until the last moment, just in time to make that exit.

This is more efficient driving, from the perspective of most drivers (I realize that you traffic researchers out there would argue that this is lousy driving and worsens traffic, and it is unsafe driving, but anyway that’s a different debate for another day).

We are building “procrastination” into the AI of self-driving cars. Rather than being the goody two shoes kind of driver, our view is that if self-driving cars are supposed to be able ultimately drive as good as a human, they should be adopting human driving techniques. The Utopians out there are going to go ape, since they believe that all AI self-driving cars will be perfectly civil and obey all laws and be sweet and kind to all other cars on the roads. Maybe. I don’t think so.

Conclusion

There are many driving situations wherein “procrastination” (the good kind) is actually handy to consider as part of the driving repertoire. Of course, it shouldn’t be used all the time.

Some want AI self-driving cars that can pass a variation of the Turing Test. This means that AI self-driving cars would be able to drive a car in the same manner that a human does. If we were standing outside and watching two cars drive along, and we couldn’t see into the cars, and we had to try and say which one was being driven by the human and which by the AI — if we couldn’t discern which was which, in a small way the AI has passed a type of Turing Test.

Today’s AI self-driving cars are programmed to drive a car like a teenage novice driver, and indeed less so since the AI of today does not have common-sense reasoning and lacks a myriad of other human-thinking qualities. This state of affairs is not going to get us to the vaunted Level 5, which is a true AI autonomous car. Using the technique of purposeful “procrastination” is one of the many ways in which an AI self-driving car can drive more like a human can.

For free podcast of this story, visit: http://ai-selfdriving-cars.libsyn.com/website

The podcasts are also available on Spotify, iTunes, iHeartRadio, etc.

More info about AI self-driving cars, see: www.ai-selfdriving-cars.guru

To follow Lance Eliot on Twitter: @LanceEliot

For his Forbes.com blog, see: https://forbes.com/sites/lanceeliot/

For his Medium blog, see: https://medium.com/@lance.eliot

For Dr. Eliot’s books, see: https://www.amazon.com/Dr.-Lance-Eliot/e/B07Q2FQ7G4

Copyright © 2019 Dr. Lance B. Eliot

Written by

Dr. Lance B. Eliot is a renowned global expert on AI, Stanford Fellow at Stanford University, was a professor at USC, headed an AI Lab, top exec at a major VC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store