Google's Gemini is...not that bad?
Google Gemini Logo
OK, I will start out by saying that I understand the reservations many people have about AI bots, and the risk of using them. They are definitely capable of giving inaccurate information and should not be relied upon as factual sources. If I read something that cites information from ChatGPT or whatever I'm not going to finish reading it (not that I've ever come across such a theoretical article).
That said, I believe that AI tools can be useful. I recently used ChatGPT to get some ideas for camera settings to replicate film looks on my smartphone, because even in Pro mode I don't have as much flexibility as I would with my regular mirrorless DSLR. I have photography books I can consult that give me information about how to manipulate things like ISO/Apeture/Shutter Speed etc. and what they do, but not only do I not have these handy "in the field", but they don't tell me how I might replicate a specific type of classic film.
I (and others I know) have also used these AI tools to generate starting points for various types of writing, or asking for quick grammar checks. Or sometimes asking for a re-write in a particular style. I don't believe any of these examples are substitutes for an actual person - I would never send an important document through an AI tool for proofreading or official translating.
I was holding out adding Gemini to my phone, mostly because I dislike how Meta has shoved its own AI tool into every aspect of their apps: Facebook, Messenger, WhatsApp. It could even be in Threads or Instagram but I don't use those enough to notice. I don't think Copilot has made its way to my Surface but if it has, I haven't noticed. What I'm saying is the part about AI bots that I dislike are when it's forced upon you by a software company trying to hitch onto the AI Train and capitalize on it.
Gemini - so far as I've seen - has been strictly opt-in. Sure, I was prompted a few times because they really, really would like you to install it. It's their new pride & joy. The other day, I finally installed it, to try it out. It seems easy enough to uninstall it and disentangle it from my phone & Google Assistant.
At first I tried a few mundane things, like checking to see if Flickr was down, asking it about outfit colour matching, if No Frills did price matching. Then I tried some other fun things like asking it to look at a picture on my phone's screen, and identify the movie (which it did successfully). I tried getting it to generate an image for a blog post but it failed pretty badly at that one.
I can type in my queries or use my voice (haven't tried that one yet). I haven't tried asking it anything where I would need actual facts, and I don't think I will. You can't beat real research.
One thing I'm not sure about is how it gathers the information. I assume that it scrapes websites, but I'm unsure if it is ignoring websites that specifically say "no you can't take information from this website". There's a whole issue about that: https://rknight.me/blog/perplexity-ai-robotstxt-and-other-questions/
Anyway, that's just my opinion of it so far after a few days. You'll notice throughout the post I used italics a lot in certain places. Some of the words I highlighted:
tools, starting points, opt-in (this was bold, not italics)
I did that for a reason - to emphasize that these AI things should be viewed as tools to help you complete a task, and are definitely not capable of replacing someone's job or workflow. I might ask an AI bot to generate a simple image for me (text on a background or something) but if I needed something that requires actual creativity that I couldn't do myself, I'd pay a graphic designer. I mean, I'd try it myself first, but would probably fall back on someone who knows what they're doing.
Reply by email
Or if you prefer, find me on Mastodon.