wgrover
Gemini wrongly called an authentic product a fake, but that was mainly because there really were typographical errors in the authentic product's ingredient list. That's more of an indictment of the manufacturer than it is Gemini.
Finally, a lot of the things that Gemini got wrong seemed to me like they could reasonably be attributed to things like optical artifacts in the photos (glare, shadows, stuff like that). Better/more photos might improve that.
All that being said, this is obviously a tiny study of a single AI tool with a single brand of cosmetic product, so it's probably premature for me to optimistic, even cautiously. I've edited the last section of the writeup accordingly.
I'm hopeful that future AI models (perhaps trained for this purpose, not a general-purpose model like I used here) might be able to identify more subtle variations in e.g. injection molding patterns, the surface finish of a pill, things that would be hopefully a lot harder for counterfeiters to fix than a typo.
Identifying fake cosmetics using artificial intelligence
You can get some of these (like the smaller TOMY robots) for pretty cheap on eBay. They're usually broken, but the innards of these robots are so interesting, just taking them apart is a learning experience. TOMY was brilliant at making seemingly sophisticated toys that were actually run by a single DC motor; all the movements, sounds, sensing etc. were implemented using gears and cam shafts and other mechanisms. A great way to learn about simple machines, and a bonus if you (or your kid) can repair them and bring a 40-year-old robot back to life.
https://www.reddit.com/r/architecture/comments/5w9ep7/cross_...
https://americanliterature.com/author/mark-twain/short-story...
https://github.com/wgrover/jamie
The name 'jamie' is in honor of Mythbuster Jamie Hyneman's meticulously labeled storage bins at M5 Industries.
The code uses a brute-force search to fit the specified text on the label using the largest possible font size.
Every email client I've used since Pine has been better at dealing with all the stuff we've bolted on to email (HTML, attachments, calendar invites...) but none of them have been as efficient at the core tasks of email as Pine was.
A side note: I think there's an unmet need for algorithms that can convert photos of these random patterns into text (or something similar) that can be stored in a database and searched quickly for matching patterns. I've tried image similarity algorithms like the ones used by e.g. Google Reverse Image Search, but they seem poorly suited for this task. I ended up writing my own crude algorithm in the paper above that converts a pattern into a set of strings, and it works OK, but surely there are better ways to do this.
[1] https://en.wikipedia.org/wiki/The_Pirates!_In_an_Adventure_w...
Random musings of a professor of Bioengineering at the University of California, Riverside. A few highlights:
- https://groverlab.org/hnbfpr/2017-12-10-csu.html - My investigation into a fictitious California university and its link to predatory academic journals.
- https://groverlab.org/hnbfpr/2019-08-19-gene-roddenberry-ucr... - Looking into filming locations for Gene Roddenberry's TV show pilot "Genesis II" that was filmed at UC Riverside in 1972.
- https://groverlab.org/hnbfpr/2019-08-06-stereo-records.html - How stereo phonograph records work.
And a fun video of Adam Savage playing with Marcus Wu's 3D-printed Curta: https://www.youtube.com/watch?v=j9uRckJLqLk