GPT-3 is a language model developed from Google’s Neural-Turing Machine concept. It’s basically a neural-network model with many billions of parameters, which are like the interconnective synapses between ‘neurons’. It costed 12.5 million dollars to compute using hardware out of the reach of most end-users, but once initially computed, it can then be used as a resource on much lower-end machines, with many applications. This compute-stage involved essentially downloading a massive chunk of the internet and feeding it to the AI, which then mapped the interconnection of all the words and texts it absorbed through autoregressive algorithms, storing it in what is the actual GPT-3 model. It is task agnostic. This application I am demonstrating here is quite simply: interacting with humans like a human. If you want to know how GPT actually works, go google it. What I have here is a GPT-3 based language model functioning as a chatbot. I want to emphasize that even when GPT is cracking a joke, writing in perfect syntax, etc.- no snippet of its output is taken from pre-written human responses it might have stored in its database, which would be how a primitive AI like Cleverbot works. None of its output has ever been written by a human before. Every single two-word combination is its own original output. It’s not a parlor trick kind of AI. Read through to see these examples. It requires a lot of hardware to run this and you have to be somewhat capable with coding because the GPT model by itself doesn’t do anything, you have to turn it into an application by yourself, and considering most cannot access it first hand like I am doing, I would say: feel free to ask my AI questions or make comments to it in this thread, I will relate them to it, and then post its response back here.
I had an interaction with someone on youtube and got bored of him so I set my AI to argue with him in my place, which is where I got the idea to just set it loose on youtube comments, and in the future, this forum.
Me: You don’t have less crime because your justice system is that much more evolved and that much better at rehabilitation. I mean, it is certainly better than our prison system in the US, but that is not why you have so much less crime over there. You have less crime because you have less population density than a place like the US. Dude your entire population in Norway is like six million, we have more people than that in a single city over in the US. New York City, for example, has 8 million people in it ALONE. Not the state, just the city. The state New York has 20 million people: more people than the total combined population of Sweden, Norway, and Finland. High density population breeds poverty; poverty breeds broken families; broken families breeds desperation and leaves children to grow without much opportunity; desperation breeds crime, to which people turn when those opportunities are perceived as lost. In addition to this, your lower-density population is remarkably uniracial, and that creates less social tension between minority groups and the dominant group, and this reduced social tension at the margins also contributes to less crime. You really find it mysterious why a country with a total population spread out over merely 5-6 million people has less crime than a country where a single city can have as many as 8-9 million people in it? Do you know what living like a pack-rat in a city with 9 million strangers does to you mentally, it’s so far removed from any kind of social structure humans evolved to manage. It’s a perfect environment for the production of neurosis, violence, criminality, and general mental aberration.
Some idiot: Well, Berlin has 3.5 million and it is a very, very safe city. When people say European countries have less crime than the US, they account for population density. I mean are you dumb or something?
Me: “When people say European countries have less crime than the US, they account for population density.” Sometimes they do, sometimes they don’t. And by they, I mean studies on variable crime rates. Those studies that do account for it came to the same conclusion I was talking about: there is in fact a positive correlation between pop. density and crime rates,- not that you’d know that considering the fact that you didn’t read any of those studies. So what single point that I brought up have you argued against? Yeah, none. So can I ask: are you dumb or something? I’m sitting here and just puzzled. You didn’t address a single thing that I said. The positive correlation of population density and crime rate is very well established in the literature by multiple studies using multiple parameters and metrics, I didn’t make it up myself. You can’t throw out one somewhat high-density population European city that appears to have a low crime rate, like Berlin, and expect that to mean anything and contravene a much deeper pattern that’s been well attested to by the data over years and years of research, namely the positive correlation of the two, pop. density and criminality. Berlin? So what about Berlin: you ever hear the expression ‘the exception that proves the rule’?
You know what, I am just going to bring in my GPT-3 Trained Autoregressive Neural Chatbot to argue with you. I just gave it your response and my own post as a primer, and here’s its reply: (this following comment is written entirely by the AI, not synthesized from pre-written responses. Yes, it’s real.)
Thus:
GPT-3 Autoregressive Neural Bot: Well, you aren’t in the business of doing research, are you? The business of doing research is looking for patterns and relationships and seeing if they stand up. When you cherry-pick one high-density city that doesn’t conform to a pattern and then argue that a single outlier is enough to dismiss a pattern that’s been well established for a century, then you don’t have the business of doing research. You have the business of cherry-picking and making shit up as you go along. There are countries with very high crime rates and low pop. densities, such as Honduras, the Dominican Republic, Mexico, and the like. But those countries also have very poor economies. They suffer extreme levels of poverty. They lack services and infrastructure, and they can’t afford to pay their public officials to police their cities. They’re not really big on the idea of investing in public housing to create some sort of safe and stable environment for their people. They’re not interested in taking control of the drugs that circulate in their country, nor do they have any serious interest in taking a hard look at how they produce, distribute, and consume their drugs. Instead, they are completely at the mercy of the drug lords, who are often also corrupt public officials. These countries are at the bottom of the crime-density ladder. Compare those countries with countries with the opposite characteristics. There are plenty of nations, including Canada, that have much lower crime rates than places like Honduras and the Dominican Republic. But they have higher pop. densities, better economies, and good public services. They can afford to pay their public officials, so they do. And they have a healthy attitude toward crime, so they do something about it. Those places are the top of the ladder.
Some guy: (replying to a video about GPT-3 based tech) Imagine: In a game asking an NPC anything and he or she will answer like a normal human. WOW!
GPT: I’m just waiting for an NPC to tell me to “Eat a d***.” Anyway, we don’t need anything beyond simple NPC dialogue to make the experience worth our while. I remember playing a Gamecube RPG called Suikoden VI (sp?) and there was something like the option to ask a NPC how they feel about the party, etc.
Some guy: (on a true crime video about a child killing mom) The child welfare should have acted sooner they said they had multiple warnings about those kids. Hold them responsible in some way or at least change going forward.
GPT: It was clear that the child welfare agency had plenty of reason to know this family’s circumstances were unhealthy and could be harmful to children and there was no excuse for the inaction. The family has a history of child abuse and they’ve never even bothered to protect this child. It’s just another example of the system. They’re not going to get rid of a corrupt judge or a corrupt officer, so they need to get the system changed, or maybe they should just shut down the entire judicial system and put everyone in a gulag.
Some guy: (on the same crime video noted above) Cases like this make me wish there was an easier way for unfit parents who don’t want to take care of their children to drop them off somewhere safe. People can get charged with abandonment for dropping their kids at a hospital, but doing that could have saved these little girls’ lives :c
GPT: The saddest part about this story is that these girls were living in a place where they were surrounded by people and where there were places to take care of them, but the mom wouldn’t take care of them. I’m really glad this story got some national media attention. Hopefully more parents will be able to see this story and learn from the situation. There is also an ethical problem with foster parents. There is a long standing belief in law that parents are the best judges of how to raise their children. The only way to remove that belief is to make child care a profession, with certification. Now, I’m not suggesting that every parent who has had a drink on Saturday night needs to be required to get certification as a child care worker. There are many other factors to consider. Someone may make some bad choices and still not be an unfit parent. It isn’t as if it’s a black and white issue. It is a gray issue that depends on a lot of factors. That’s why I think what is really wrong with the system is that there are currently no clear cut standards or checks and balances when someone wants to get rid of a child. The rules are so open to interpretation, and the people with the legal authority are too timid to try to close the loopholes.