02C02: AIs Everywhere

Thursday Keto Pasta

Metro Manila Community Quarantine - Day 1006
Reading Target: 43 books left

The release of the Lensa app has become a rallying point for discussing the implications of so-called "AI art", or whatever you want to call the algorithmically-derived images that are largely created without human input. This is not the first machine learning-powered tool that has gotten people to question what is an appropriate use for this technology but its certainly the most omnipresent with so many people flooding their social feeds with iteration upon iteration of profile photos or other images made in the styles of other artwork seen elsewhere.

On the one hand, there is a valid concern about how many of these AI tools have been trained. We don't exactly have clear copyright protection to prevent one's creations from being used by a machine learning system and it's sort of a gray area for how that goes. But ethical sourcing of modeling data aside, there are serious questions about the technology in itself and what it does.

There'll always be a bit of a premium put on things made by people - we're just one of those societies. And we've seen different areas of our lives affected by industrialization, automation, and technology in general. And every time we're faced with these questions about what people should do versus what we should allow machines to handle.

A lot of times, we'd use human creativity as evidence of sentience or at least to distinguish humans from animals and all that. And I can understand how that adds some extra layers of discomfort when we have computer programs that can seemingly create things. Then again, these tools only iterate and don't necessarily innovate. They look for patterns in other pieces of art and used that as a basis for generating images of a similar style. This is not about computers becoming creative but maybe the algorithms are detecting patterns across art styles that we haven't picked up on ourselves.

It's not like this is the only space where machine-learning tools have started to make serious headway. We've seen AI-powered translation tools and others that can even write copy. Algorithms help ad systems better target users and they help recommend what movies and shows we should watch next on whichever streaming we're using. There are a lot of areas of our modern lives that have already been impacted by machine-learning solutions to some extent, but they've largely been more behind the scenes. 

As always, technology continues to advance at a faster pace than people are accustomed to. And now we're already monetizing something that people will probably need more time to figure out how we feel about them. And there are even more questions that are bound to come up.

I don't really have answers either. I just recognize that the technologies in themselves were probably inevitable given the way things have been developing. I agree that developers need to get the appropriate permissions/rights for their data sets for their machine-learning efforts. But beyond that, we'll have to see how things continue to grow, develop and change. We've barely scratched the surface of what such technologies can achieve.