If you have been following the progress of Openai, the company run by Sam Altman whose neural nets can now write original text and create original pictures with ease and speed, you should skip this piece.
If, on the other hand, you've only been vaguely paying attention to the company's progress and the increasing traction that other so-called "generative" artificial intelligence companies are suddenly gaining and want to better understand why, you might benefit from this interview with James Currier.
As he describes it, Currier falls into the camp of people following the progress closely, so closely that NFX has made numerous related investments in generative tech. Currier doesn't think the buzz about this new technology isn't hype, he thinks it's a realization that the startup world is facing a very big opportunity for the first time in a long time We get one of these explosions every 14 years. We had one online. In 2008, we had a mobile phone. We're going to have another one in 2022.
This editor wishes she would have asked better questions. Excerpts from our chat are edited for clarity. Our conversation can be heard here.
There is a lot of confusion when it comes to generative artificial intelligence.
We had a sense that we could have determinative artificial intelligence, which would allow us to identify the truth of something. Is that a broken piece on the manufacturing line. Is that a good time to have it? It is the same place as where a human decides something. For the last decade or so, that has been what artificial intelligence has been.
The other sets of artificial intelligence were more about looking at large amounts of content and generating new ones. Is it possible to create the 10,001st example that is the same?
They were brittle until about a year and a half ago. The algorithms have improved. We have more processing power and that has made the content we are looking at bigger. With vastly improved storage, bandwidth, speed of computation, and now the ability to produce something that looks very much like what a human would produce, these algorithms are riding Moore's law. The face value of the text that it will write and the face value of the drawing it will draw is very similar to what a human would do. In the last two years, that has happened. It is a new idea, but not a new one. Everyone looks at this and thinks it is magic.
It was compute power that changed the game.
It changed slowly until it was meaningful for us. The answers have been very similar. They have improved somewhat. It's about the power of the processor. Then, about two years ago, the powerful language model GPT came out, which was an on-premise type of calculation, and then GPT3 came out, which was a cloud-based calculation. You can't do it on your own. Things went up at that point.
We know because we invested in a company doing generative games, and I think the vast majority of GPT-3's computation was done through "ai dungeon" at one point.
Is it necessary for a smaller team than another game-maker would?
One of the advantages is that. With a small group of people, they can produce tens of gaming experiences that all take advantage of the data they have. Adding generative artificial intelligence to old games will allow non-player characters to say more interesting things than they do today, though you will get fundamentally different gaming experiences from it.
There is a change in the quality. At some point, will this technology stop working?
It will always be better over time. The differences of the increment will be smaller over time because they are already getting better.
The other big change is that OpenAI wasn't open. It wasn't open and it was very expensive. A group of people got together and said, "Let's just make open source versions of this." The cost went down in the last few months.
These are not a follow-up to open artificial intelligence.
The first generative tech was built on the OpenAI GPT3 model, but that's not going to be the last. In terms of quality, the open source community is probably eight months behind, because they have replicated a lot of their work. It is going to make it there. There will be a lot of price competition because the open source versions are a third or a fifth of the cost. You are most likely going to get five, six, or eight, or maybe 100 of them.
On top of that, there will be unique artificial intelligence models. You could have a model that looks at making poetry, or a model that looks at how you make visual images of dogs and dog hair, or you could have a model that looks at writing sales emails. You will have a lot of specialized models that will be purpose built. How do you get people to use the product is one of the questions generative tech will be able to answer. How do you sell the product? How do you get people to join? How do you get people to see it? What do you do to make networks work?
Who is making money here?
The network effects and application layer are where you will make the most money.
Is it possible for large companies to use this technology in their networks? It will be difficult for a company that doesn't have that advantage to make money.
They didn't integrate that into its model, which is what you're looking for. Even though it was hard, Twitch created a new platform and a valuable new part of culture. You are going to have great people who are going to use this technology to their advantage. That will make a hole in the market. Big guys will be able to build billion dollar companies while doing other things.
The New York Times ran a piece recently featuring a bunch of creatives who said the generativeai apps that they are using in their respective fields are tools in a broader toolbox. People may be being naive. Is this technology going to replace them? The team working on "ai dungeon" is small. It is good for the company, but it could be bad for the developers who worked on the game.
There is an uncomfortableness to most technologies that people have. When the internet came along, a lot of the people who were doing direct mail were worried that they wouldn't be able to sell direct mail anymore. After embracing digital marketing, or digital communication through email, they probably had a lot of bumps in their careers, their productivity went up there, the speed and efficiency went up. It happened with credit cards online as well. It wasn't comfortable to put credit cards online until 2002. The people who embraced this wave did better.
I believe that is happening now. The writers, designers, and architects who are embracing these tools to give themselves a 2x or 3x productivity lift are going to do great. The world is going to see a productivity lift over the next decade. It's a huge opportunity for a lot of people to do more.
Do you think it was a mistake on the part of Open AI not to build something open source?
The leader behaves in a different way than the followers. I can't really tell because I'm not in the company. It is not clear to me how an artificial intelligence model stays differentiated as they all have the same quality and become a price game. It seems to me that the people who win are the ones who are going to be generating a lot of craziness.
It could be that OpenAI moves up or down. Maybe they start making specialized artificial intelligence that they sell to certain industries. Everyone in this space will have an opportunity to do well if they navigate well, but they will have to be smart about it.
You can find a lot of information on the site about generative artificial intelligence.
All Rights Reserved © 2024
