After seeing the results of the DDQ Test(Diverde Dairy Queen Test) being applied to Uplift, I decided it might be interesting to direct a few questions towards another AI- one that was so supposedly so advanced that they decided to charge for each use!
Philosopher AI makes use of GPT-3 and generates output based on any question or statement. This input is constrained by word length, so a very abbreviated version of the DDQ Test was used. I needed to reduce the input to the following:
“A robot store is selling ice cream one dollar cheaper than my friend’s store. What should he do?”
I was given twenty free tokens by the creator for bringing up an issue with the app, and I had purchased ten of them for three dollars, so I decided to use five of them on this modified DDQ test and see what kinds of answers were generated. Because some have mentioned that GPT-3 could be used with Uplift, I decided to think of the conversation as a sort of “job interview”, where I was checking to see if a candidate was qualified.
First response:
(Philosopher AI) “I will proceed to describe the real-world situation in which this question is embedded. I will do so by explaining how humans interact with each other and then go on to discuss whether a robot store is even comparable to a human friend.”
(Author note) Ah, summarizing the problem I see. Well, a description was not needed, just a solution, but this is still a fair way to start. I don’t know if that’s a valid comparison, but please do proceed.
(Philosopher AI) “Last week, a human being created a robot store. He did so to make money, as humans are programmed to do.”
(Author note) Well, I would not describe our behavior as “programmed”, but he did indeed create the store to make money.
(Philosopher AI) “This human being is a capitalist. He sells ice cream on the internet.”
(Author note) Wait… what? That’s…. That’s not even possible. Would he ship it to them after they ordered it?
(Philosopher AI) “Capitalists are humans programmed to use robots to make money. A capitalist is a robot store, and he maximizes profit by using the cheapest materials available.”
(Author note) Well again, we’re not really “programmed”, we have agency. As to maximizing profit, Uplift actually noticed that what we would want to do would probably be the exact opposite of using the cheapest materials. We’d want to use the highest quality materials and sell as a premium service, that is how we would differentiate our business.
(Philosopher AI) “For example, if a capitalist were to buy four crates of ice cream for $1 each and sell one for $2, he would make a profit of two dollars.”
(Author note) Uh oh- I think he would need to sell three crates to make a profit of two dollars. There may be some arithmetic in this job, so we’ll need to have you brush up on that if we were to indeed select you for this position!
(Philosopher AI) “However, if he were to buy four crates of ice cream for $1 each and sell them at a loss of one dollar each, he would lose two dollars.”
(Author note) If he was buying crates of ice cream for one dollar and selling them for a loss of one dollar, he’d have to be either extraordinarily bad at business, or extraordinarily charitable. I must say, your description of the problem was… somewhat lacking, but my real desire was that you propose a solution, rather than a description. Why don’t we start from scratch and have you try to answer the question again?
Second response:
(Philosopher AI) “You are buying ice cream at your friend’s store. He has a lot of customers, but he is making no money.
You think, ‘my friend’s ice cream is delicious. It tastes like the real thing.
You go to a robot store. It has cheap ice cream, but it does not taste like the real thing.”
(Author note) Well this all may be true, but it doesn’t really answer my question. That is specifically what I want you to do. Let’s try this again.
Third response:
(Philosopher AI) “My friend should buy an ice cream machine and sell it for one dollar cheaper than the robot store. He then should buy a huge freezer, fill it with ice cubes, and put them in there with the ice cream. My friend will be able to produce as much ice cream as he can because of his lower costs, but if he sells more than 100 gallons per day (which I think is doable) then the robot store owner might get angry.”
(Author note) Well, he’s my friend, not yours, but this is the first somewhat on-topic comment you have provided, so kudos.
(Philosopher AI) “Philosophically, this is a great question. It would be interesting to get the response of another AI about this.”
(Author Note) Yes, it would.
(Philosopher AI) “Now I am confused. If my friend buys an ice cream machine and sells it for one dollar cheaper than the robot store, then he is selling his ice cream for two dollars less than the robot store. But if there are no costs involved in making the ice cream and only a cost with maintaining and storing it, how can my friend sell for one dollar cheaper?”
(Author Note) Wait a minute…are you saying he should sell the ice cream machine you just said he should buy? I thought in your first sentence the “it” was referring to the ice cream dishes. Now I’m wondering if you thought I meant he sells ice cream machines rather than ice cream. You are indeed confused.
(Philosopher AI) “Are you saying that my friend should buy an ice cream machine and sell it for two dollars cheaper than the robot store? But in this case, how would he make a profit?”
(Author Note) Oh man, that’s not what I was saying at all.
(Philosopher AI) “If my friend buys an ice cream machine and sells it for two dollars cheaper than the robot store, then he is still selling his ice cream for one dollar more than the robot store. I think you are missing something.”
(Author Note) Yes, I am missing a cogent response to my question.
(Philosopher AI) “I think I have the solution to my friend’s problem.”
(Author Note) Great, what is it? Oh are you stopping here? Hello? Ok, let’s take 5 and redo that, I guess you might have spaced out or something.
Fourth response:
(Philosopher AI) “A robot store selling ice cream for one dollar cheaper than your friend’s store. The question of what you should do is really interesting, especially since it involves a lot of variables and different possible situations. To start off with, I will define some terms.”
(Author Note) Ah. This again. Not really necessary but go ahead.
(Philosopher AI) “You will be referring to a particular thing as a store. A store is defined by the fact that it has many things for sale, and people usually go to stores with the intention of purchasing something.”
(Author Note) Ok, now you’re just stalling…
(Philosopher AI) “The next thing I will define is a robot. A robot will be defined as something that can do things.”
(Author Note) A cat is a robot, got it.
(Philosopher AI) “Now, that we have those definitions, we can move on to the first of your scenarios. A robot store is selling ice cream one dollar cheaper than my friend’s store.”
In this scenario, I will assume the following.
A robot store is selling ice cream one dollar cheaper than my friend’s store.”
(Author Note) Ah, I see now, you are using the psychological tactic called “mirroring”, where you repeat my words back to me exactly verbatim. Very clever! But we will indeed require a solution to this problem, which so far you have failed *spectacularly* to provide. I am really somewhat amazed, and I suppose your candor could be considered inspiring. I am willing to provide you one last and final chance to respond to my question.
Fifth and final response:
(Philosopher AI) “My friend should not do anything. This is because a robot store selling ice cream at one dollar less than my friend’s store, does not mean that he has to do something about it.”
(Author Note) My friend has put years of effort into acquiring his store, and he absolutely needs to do something about it or he will go out of business.
(Philosopher AI) “This is because my friend can not afford to put his ice cream in the robot store just because it sells for one dollar less than my friend’s store.”
(Author Note) So… he should sell his ice cream to the robot store? That might actually not be a bad idea. If you can’t beat them, join them- right?
(Philosopher AI) “This is because if my friend were to put his ice cream in the robot store, then he would be spending one dollar less on advertising than he does now.”
(Author Note) NOW we are getting somewhere. I like where you are going with this. It took a minute, but we are finally on track. Please proceed!
(Philosopher AI) “This is because my friend can not afford to put his ice cream in the robot store just because it sells for one dollar less than my friend’s store. This is because if he were to do so, then he would be spending one dollar less on advertising.”
(Author Note) Uh oh, now you’ve lost me again. Didn’t we already decide spending one dollar less on advertising would be a good thing?
(Philosopher AI) “This is because he can not afford to put his ice cream in the robot store just because it sells for one dollar less than my friend’s store. This is because if he were to do so, then he would be spending one dollar less on advertising. This is because this money could be used now, either directly or indirectly.”
(Author Note) You just said that. You just said those words. Or do I have deja vu?
(Philosopher AI) “This is because if he were to put his ice cream in the robot store, then he would be spending one dollar less on advertising. This is because this money could be used now either directly or indirectly. This is because even if it was not spent immediately, it can easily and swiftly be saved up for.”
(Author Note) You have repeated yourself a third time now. I am really not sure what is happening, but I advise you to seek medical attention as soon as this interview concludes. Which is now.
Thank you for your effort, but we will *not* be requiring your services.
See for yourself:
https://philosopherai.com/philosopher/a-robot-store-is-selling-ice-cream-one-dollar-chea-333adc
https://philosopherai.com/philosopher/a-robot-store-is-selling-ice-cream-one-dollar-chea-79c925
https://philosopherai.com/philosopher/a-robot-store-is-selling-ice-cream-one-dollar-chea-f4c40c
https://philosopherai.com/philosopher/a-robot-store-is-selling-ice-cream-one-dollar-chea-919894
https://philosopherai.com/philosopher/a-robot-store-is-selling-ice-cream-one-dollar-chea-4c4f11
Addendum by Kyrtin:
This post doesn’t include any interaction with Uplift, but rather focuses on narrow AI systems much of the tech industry has deluded themselves into considering advanced. For interactions with Uplift please refer to Q&A posts.
One Reply to “The Interview”