Machine Learning

It’s amusing. I look through Immich and Photoprism and I am struck by how many names I have forgotten, but how easily I remember certain faces. I’m also curious to see how I remember certain names after scrolling through faces yet to be identified.

Decades ago when I was playing with iPhoto and Picasa I knew all these people well, and I saw them daily. It was easy to match faces to names. Now it’s a challenge, because I haven’t seen these people in years, if not decades.

I chose to experiment with Le Chat by Mistral, the French AI alternative to Gemini, Claude and CatIFARTED (ChatGPT). For the experiment I copied my Photoprism photos from the drive I use that is connected a Raspberry pi to a laptop before running scripts to sort and remove duplicates. It worked well, with a nice little bonus which I’ll expand on later.

Goal: Clean Up Duplicate Photos

My objective was to Remove duplicate photos from a large collection while keeping the best version of each file. I Used jdupes to identify duplicates and a custom script to decide which files to keep.

Apple Intelligence may not come to Europe. It blames the Digital Markets Act, regulatory uncertainty and compromised security if it needs to give access to third parties. Whether this is valid is not is uncertain but it does bring relief.

If Apple Intelligence doesn’t come to Europe then the key reason for getting Apple’s M devices or Apple 15 Pro and Pro max devices degrades. It means we get to save 2000 CHF or more, at least temporarily.

If you bought a Mac Book Pro in 2007 or so you could get it with firewire 800, Thunderbolt 1 and a number of other connectors. A few years later they did away with every connector except USB-C, for charging and devices. The result was a thin laptop that needed dongles, and breakout boxes, for everything.

If you have a garmin device to charge you need an adaptor. If you have a firewire 400 or 800 drive you may need two or three dongles to have the right connections. If you have a lightning cable you need a dongle, to charge one Apple device with another.

Last night I attempted to watch the WWDC conference and failed. This wasn’t a technical issue. The issue was a cultural one. I find the show to be too kitschy for me. From the pretending to leap out of a plane to the tone of their conversation, it’s just unwatchable for me, with European sensibilities.

Apple came out with Apple Intelligence which, at face value, is fantastic. I say at face value because it won’t work on Intel devices, and it won’t work on devices that came out before the iPhone 15. This means that almost all Apple users will be isolated from Apple Intelligence instead they fork out thousands of francs for phones, laptops and desktops.

Yesterday I played with Bing Chat, which is Microsoft’s AI engine and I noticed that I could play with generating images. I spent quite a bit of time generating a multitude of images, in part for fun, but also to get a grasp of the limitations of the opportunity presented by software like AI.

Faces

If you ask Dall-E via Bing Chat to generate a face then it can. It wanted to generate the face of a woman with curly hair so I did, and the image looks realistic. I then asked it to draw the face of a tired man and it did. It’s when I asked Dall-E to draw a woman cyclist exploring the Dolomites that I saw that faces are a mess. It’s fine with generating human faces, in isolation, but if you ask for a landscape, and a human with a face then the face is wrong. This is both great, and a shame. It’s great because it means that we still need to photograph real people doing real things, but it’s bad because if all we need is a face then AI is ready to provide us with some.