As someone who has started messing around with different open TensorFlow models and NLP algorithms I believe there’s is a lot of exciting things that AI/ML will unlock, although there is far too much hype right now.<p>Still despite the hype there are a lot product opportunities and we are only beginning to scratch the surface thanks to these phones getting better and generally large scale compute becoming easier to access. However, I disagree with the author that we can’t go vertical for now and have an app for each little advancement to take advantage of the piece meal innovations in this space. I think that’s exactly what we should be doing ala Shazam, as he mentions. I think product discoverability and growth is a whole other problem and I don’t think one needs a fully integrated magic AI/ML camera on day 1 to make an impact and have a great product, old techniques coupled with some of the newer open source models/papers these big companies are putting out can work wonders.<p>As a little example of this, and some self promotion, I recently built a cross platform app that uses some AI/ML techniques to read any article to you from the web, forgive me for being a little self promotional but if your interested you can check it out here: <a href="https://articulu.com" rel="nofollow">https://articulu.com</a>
I have to cop to being the philistine who welcomes this, while simultaneously feeling a little uncomfortable with it. I bought an entry-level DSLR many years ago, in the naive expectation that if I studied and practiced I would be able to capture something of the really amazing experiences I had traveling, camping, and watching my city change and grow. After a lot of frustration, reading, and talking with more experienced amateur photographers, I realized that's not what photography is about. Noticing that something looks amazing or beautiful to the human eye and noticing that there's an amazing or beautiful photograph to be taken are completely different things, and the best photograph will be different from what you see with the naked eye. Instead of naively trying to capture my experiences, I had to learn to "see like a camera" and see the possible images that could be created via lens choice, camera settings, and post-processing.<p>Not only that, it's hard work. After talking to some better photographers about their photos I realized the experience they were having when they captured a great photo was the experience of working hard at capturing that photo. That was the only real experience they ever captured, and they did not expect there to be any trace of it in the photograph itself. In other words, my dream of capturing my experiences was a naive fantasy, and there were only two rewards: loving the process, and occasionally creating a beautiful photograph.<p>Sounds a lot like programming, doesn't it! When I bought my DSLR I was basically like the guy who decides to write an app because it would be cool to have the app and cool to have written the app, without realizing that the experience of writing the app far outweighs whatever pleasure might come from the result. If you don't derive enjoyment and pride from the process, there's no way that any result will repay the effort.<p>And I was naive enough to think I could co-opt this hard artistic work to document the experiences I had as a non-photographer! That would be next level shit. I would have to be so good at noticing and capitalizing on the possibility of beautiful photographs that I could pick and choose the ones that happened to coincidentally reflect my non-photography-mediated experience. And that means I'd have to attend to the photographic possibilities <i>at the same time</i> that I was attending to my own experience <i>that was somehow separate from photography even thought my photographic skill was actively engaged</i>. Impossible, at least at the skill level I felt I could realistically aspire to.<p><i>But</i> if the photographic expertise and cognition were in a piece of tech that I could carry with me, the whole idea would make sense again. My vacation photos would be like my wedding photos: a big stack of expertly shot images from which I could choose the ones that captured my experience best. Whoah. That's pretty much what I was looking for in the first place. I just wanted to share and remember.<p>Go ahead and embed this in my contact lenses, please! As long as I can adjust my preferences to get somewhat realistic images. I don't need ML giving me a chiseled jawline and making all my sunsets hot pink.
This analysis is late to the game. Software and AI has been eating photography for some time now.<p>Do yourself a favor: browse through any photography forum and listen in on the discussions.