LTP 113: AI Part 2 — Pros & Cons with Antonio Rosario 1

Panel

• Bart Busschots (host) – @bbusschots – Flickr
• Antonio Rosario from the Street Shots Podcast & The Unusual Collective – @amrosario – Flickr

In part 1 Bart described AI’s evolution from a photpgraher’s point of view — first came classifiers (this is a photo of a dog and a tree), then came smart processing like computational photography on smart phones, auto edit buttons that actually work, and advanced editing tools like content-aware fill, and now, here comes Generative AI, the ability to generate images from scratch using some kind of prompt. Let me tell you what’s in the shot, then let me tweak it for you, and now let me create it for you! In this second part, Bart is joined by Antonio Rosario to talk through the pros and cons of each of these kinds of AI for photographers. This then sets up a follow-on conversation on Antonio’s Street Shots Podcast where Bart joins Antonio to talk through some of the artists, moral, and ethical implications of all this.

While this podcast is free for you to enjoy, it’s not free for Bart to create. Please consider supporting the show by becoming a patron on Patreon.

Reminder – you can submit questions for future Q & A shows at http://lets-talk.ie/photoq

Classifiers

• Pros
  • Easier searching of images without metadata
  • Accessibility for the visually impaired
• Cons
  • Atypical views of things are most likely not to get matched, and hence, get lost — biases in favour of samey shots and against more creative shots
  • Over confidence in the algorithms could discourage manual tagging, descriptions, etc, reducing the quality of metadata overall

ML Editors

• Pros
  • Spectacular improvements in image quality from small sensors, i.e. why phone cameras don’t suck anymore!
  • Democratisation of image editing — regular folks get better images!
  • All editors of all levels get freed from drudgery with powerful automations like content aware fill
• Cons
  • The algorithms can end up defining the look of millions of people’s photos — the average rises, but the peaks fall too
  • It can be very hard to correct images when computational photography gets it wrong, much ML is now so engrained in the processes that it can’t be turned off!

Generative AI

• Pros
  • Lets imaginative people who are not skilled image creators create amazing visual art
  • Can stimulate creativity
  • Democratises basic illustrations for articles, presentations, etc.
  • Allows the creation of otherwise impossible things
• Cons
  • Could put photographers out of a job
  • Could crowd out human creativity
  • The legal footing is simple in limbo — using generated images opens you up to real legal jeopardy because absolutely no one has any idea how the courts are going to treat this!