What We Talk About When We Talk About Bots

I’m Kelly Vero, your friendly neighbourhood robot lover. Today, I want to take us beyond the famous Raymond Carver short story collection, the chicken sludge for the soul, from 1981. 1981, of course, was a very good year for science fiction.

It was a typical Saturday afternoon when Jack sat down at his computer to catch up with his old friend Tom. Tom had always been fascinated by artificial intelligence, and he was eager to tell Jack about the new chatbot he had been working on.

"It's called ChatGPT," Tom explained. "It uses natural language processing and machine learning to converse with users in a way that feels natural and intuitive."

Jack was intrigued. He had heard of chatbots before, but he had never used one that was as advanced as ChatGPT sounded.

"How does it work?" Jack asked.

"It's actually quite simple," Tom replied. "You just type in your message and ChatGPT generates a response based on its understanding of language and context. And the best part is that it learns and adapts over time, so it only gets better as you use it."

I asked ChatGPT to start a love story about ChatGPT in the style of Raymond Carver. Could I be more meta? Everybody loves ChatGPT, don’t they? Not forgetting Midjourney or Stable Diffusion. Dall-E? Yes, you lazy gits. But also, everyone loves a good love story - especially at this time of year. Here’s mine.

Why Don't You AI?

Mostly because you don’t know what it is. So let's start at the beginning with ML. Machine Learning is a field of inquiry that understands a series of questions or commands to improve the performance of something. It’s part of the artificial intelligence fam, but in my opinion, much more of an oracle or database of information storage like a data lake, waiting to be channelled into whatever you want or need.

What you are supposed to do with ML is ask questions. Any questions or prompts. From what is Kelly’s favourite colour (it’s not actually pink) to What is the secret of the Grail, who does it serve? (only Sir Perceval knew the answer before he was found, eyes pecked out in a forest). What pops out at the other end of ML is entirely up to what you design for it, what your outcomes are - we call this AI.

I like to simplify this further when I talk about AI and ML thus: ML asks the questions, and AI writes the answers. These two work together because they have to, they're stuck together for eternity. Therefore, when you open your chat GPT, what you're effectively getting is a bunch of answers, no wait, the entire world of answers, that relates to whatever it is that you want to write or know.

But I have some beef. ML and AI is fascinating, of course it is, but this ChatGPT thing is a bit fascinating, frustrating, and cautionable. It's fascinating because we have made an entire life's work out of creating information. From being born, to AncestryDNA to tax returns. You may turn your nose up at your Fitbit or the GPS locator inside your mobile phone, but this is an information creator. Your footprint from birth to death goes somewhere, usually a database, and though it may not be interesting to everyone, it’s interesting to someone or something. For some websites and social media platforms, they might use this information very wisely to tell you about stuff, and that’s frustrating. The cautionary aspect of how ChatGPT is evolving, is that it's using this information to make you. Or, an unconscious version of you to some extent. Don’t believe me? Read my paragraph about how ML works. You are essentially ML. Let’s at least argue that by asking ChatGPT to create a Raymond Carver inspired short tale about ChatGPT, we have created a query about Raymond Carver and his style of telling stories. Now let’s replace Raymond Carver with Kelly Vero. ChatGPT wants to know everything about how I write before it can produce anything in my style of writing. Therefore, ChatGPT could as easily be me. You’ve got to want bitchy, salty prose though. You’ve got to really want it.

Mr ML and Miss AI

Can we therefore conclude that ChatGPT could eventually be used to build the most perfect picture of you (or me!) as an instrumental or distanced body and mind? Why not? All of my information already exists in databases and engines worldwide because I wrote the query to the ML by being alive. I also wrote: write this in the style of Kelly Vero, for this Metacrun.ch article, but I’m simply not famous enough, yet. You can be afraid of AI if you want to be, but I think the lid on Pandora's box should have been replaced when the hinges fell off 20 years ago. Remember that ChatGPT is an open system for learning language (and much more). AI is so much more than that half-Kubrick, half-Spielberg monster, and yet the form of AI contains both in equal merits. Artificial Intelligence is completely and utterly malleable and tactile, it's durable and visceral: it is exactly as you want to use it, nothing more, nothing less.

Created by the non-profit OpenAI, ChatGPT’s story begins in December 2015. “Sam Altman, Elon Musk, and other investors announced the formation of OpenAI and pledged over US$1 billion to the venture. The organization stated it would "freely collaborate" with other institutions and researchers by making its patents and research open to the public.” Open.

I love things in this web3 age that are open and transparent. When everyone is free to use tools, everybody gets control. Everyone is in the driver's seat of knowledge. That’s powerful, isn’t it? It's also recyclable, because don't forget that information creates information. So when AI is developing and creating the answers to the questions that machine learning asks; AI is also capable of creating new questions for ML.

This might be the thing that keeps me awake at night actually. You know, AI isn't wasteful. It doesn’t have to be dangerous. As meatsacks, we waste so much of what we create. We leave crap behind on servers, we clog up drives with careless memories, and we don’t move forward. We are the vanguards of nostalgia.

If I had a wish for ChatGPT or OpenAI generally, it would be to destroy as it creates, because we can’t let you do that, Dave. We want to control everything, why on Earth would we do a Frank and Dave and destroy everything that we’ve created? What we're doing really with ChatGPT, is just creating more of an opportunity for us to be able to build more servers for more information for HAL, ChatGPT, to read. We’re effectively allowing ChatGPT to the untidiest bedroom in the world, where there are rotten bananas under the bed. That stupid query that you created on a whim in Midjourney, DALL-E, or Stable Diffusion, what are you actually going to do with it?

Viewfinder Into The Future

You are creating crap. No, wait. You are asking bots to create crap. What are we going to do with it? All I've seen so far with Midjourney is people putting stuff on Twitter and LinkedIn, what's the point? It's not exactly going to change the world. It's a crap generator. I read someone extolling the virtues of Midjourney, DALL-E, and Stable Diffusion by saying that Leonardo da Vinci or even Pablo Picasso would have lost themselves in AI art. Really? Really? How? It is the end of times if we start thinking this way.

Creativity comes from within, from us. People go to university or art schools to study this stuff, on both sides of the canvas: from either enjoying what it is that they view to creating something that everybody should enjoy. I don't really see that happening with Midjourney and its contemporaries.

The best use case for these AI art engines (when the dust settles) is possibly something akin to procedural art environment generation for game and metaverse development. That will enable studios to get to market quickly, and well, because I need more Final Fantasy VII spin-offs than Tetsuya Nomura can currently provide. But what I don't really see is that we will be going to museums of the future, and enjoying some old shit that someone created whilst drunk during their office Christmas party because no one takes their trousers down to photocopy their arses these days. It’s inappropes.

After The Algorithm

ChatGPT should enable us to create more discoverability through language, culture, inclusion, and accessibility. Remember Yahoo! Babelfish? From 1997 to today, AI has been learning all about language. I have learned to speak Japanese, Maltese, Italian, and even Welsh since 1997 so imagine what AI has been doing with my throwaway chat-up lines? Siri doesn’t listen to everything you say anymore, however. So I’m safe to say “Ich würde gerne mal mit dir frühstücken. Darf ich dich zum Abendessen einladen?” [leave your best vomit gif in the comments]. Let’s be serious, Duolingo has been a runaway success over the last few years. We are a human race, we need social elements and knowledge to survive as well as food. Our brains eat through what we say, hear, write and see. That sounds like something Richard Matheson might write, but you get the idea.

I have talked about the opportunities of accessibility and inclusivity extensively through the development of VR and AR, but I do think that there's definitely room to be able to utilise GPT (it stands for Generative Pre-trained Transformer) and NLP (Natural Language Processing) to enable us to encompass much more accessibility. That way, software will be the ultimate adaptive device - but we have to let it in, trust it, and we have to use it. A lot. The questions have to be asked, and those answers have to guide us.

A couple of years ago, I was fortunate to invest in, work on, and later buy, a GPT-3 product, which originally for gamers, allowed them to connect their gameplay progress to a knowledge base. It was like a huge walkthrough that you didn’t need to endlessly scroll through. What we didn’t know at the time was how useful this product could be for enabling people who had visual impairments to be able to truly feel and see and understand how the environment worked inside the game, from checkpoints to game design. If you haven’t heard about the great work that Blind Burners do to help developers create accessible VR/AR and much more, you should check them out.

So, what we talk about when we talk about bots is us. What we are prepared to give over to allow machines to understand is probably already transacted upon millions of times, and what we are prepared to pull from that data lake is as far away from Megatron as you can imagine. Robots are not taking over the world. We are. Our dreams, our wishes, and even our deepest darkest secrets are a mere click away from us creating the vision of ourselves in the future. Is that so monstrous?