Are we Wasting the Potential of GPT-3 and AI Technologies?

tech bros gonna tech bro on the internet

A few segments I see tremendous amounts of innovation:

1. Using AI to generate Anime Sequences

   

2. Making a copywriting startup through GPT-3
   eg. copy.ai, copysmith, etc.

3. Using GPT-3 as some kind of AI girlfriend
   eg. replika, or whatever this post was:

   

… and maybe I’m spending too much time either on reddit or the OpenAI GPT-3 slack group and sampling innovative segments incorrectly.

But I’ve seen incredible amounts of development and even business competition over these areas compared to others.

The copywriting startups I can understand - GPT-3 is a great solution to empower copywriting, it’s a huge market, and they probably have their own differentiation between themselves (that I don't know about). But is working for interns in marketing departments the best use case for breakthrough AI advances like GPT-3? Are we incapable of imagining something greater we could do with large scale language models and their zero shot capabilities? How many more copywriting GPT-3 startups do we need? Or are we just going to keep beating this dead horse until GPT-3 writes all things on the internet and then summarizes its own writing for people on the receiving end who are not even paying attention anymore.

At the same time, maybe, people who love Anime are just more passionate about Anime and as a result they are able to form these focused groups and self organize to see meaningful results. But if this was true, why don’t other subreddits online self organize and pool knowledge/resources to make meaningful contributions to space exploration, new kinds of manufacturing, and climate change?

In some ways, this collective organization in a positive direction is already being done. I look at the EleutherAI group and am so blown away by the rate at which they are executing and the potential they have to change the status quo, to liberate large scale language models and democratize them for the people. But at the same time, will we just use these large scale language and multimodal models to just make more anime? Or to write higher converting marketing copy so we can sell more subscription bundles for loot boxes of stuff we don’t need?

I don’t want to bash any of these interests in particular because I do think it’s wrong to make fun of peoples’ interests. I have my own quirky interests too and would take it personally if somebody called me out on them. Also, I’ve recently got into anime myself to some extent. In fact, I just finished watching Akira last night and may even draw a future story board myself based on it for a video on my YouTube channel. So, yes, that makes me a hypocrite.

But really, I think it is worth asking ourselves as AI enthusiasts on a “meta” innovation level. Yes, OpenAI is focused on the safe deployment of AI models like GPT-3 and eventually DALL-E. But are we actually using these technologies effectively? How are we using GPT-3 and why aren’t we also optimizing their use cases at a meta level? Are we finding the best use cases for GPT-3 to advance our society forward or are we getting too side tracked by other causes which have potential market demand or serious fanatics on Discord/Reddit, but no real altruistic benefit. What segments, should we as a collective, choose to innovate on that would be maximally beneficial for our species? What’s stopping OpenAI from requesting areas they’d love to see GPT-3 be used for in society as a whole?

In my opinion, OpenAI already does this to some extent, I would argue the GPT-3 Beta Application process prioritizes commercial applications more over other ones. In fact, I made a whole video on it which has helped a lot of people get access by pitching commercial use cases over hobbyist ones. But, why not take it a step further? In the same way Apple enforces tight control over their applications in the app store, what’s stopping OpenAI from demanding more altruistic applications from their API partners as well? Or even both commercial + altruistic within targeted segments that could benefit society.

Is this just a reflection of the tech industry? Are we doomed to make juice pressing machines for the ultra elite or delivery apps for people who don’t ever want to leave the house to buy a burrito or something.

Are we observing a limitation of human nature? Some intersection of our lack of empathy, dudes on the internet self-organizing, or is this something else? You see, I would argue this is actually something far worse - a failure of imagination - and this could very well be our ultimate demise. We should change now while we can, and take any action at our disposal to create the type of futuristic, equitable world we want to live in. So, yes, change course now, but for the better.