I am planning a Euro trip with several stops. Most is done, but I was researching 3 cities for sightseeing. I started asking Claude AI for suggestions which led to me asking other questions and him(???) offering additional suggestions and asking me questions. I now have itineraries, instructions for transit (trams, ferries, taxi/rideshare, daily transport pass) suggested food stops, suggested timing for our group of 6. As the disclaimer says, there may be discrepancies and double check everything, but I feel I made big strides today with Claude.
What personal projects have you asked AI for help?
Gemini just helped me create a poster for an event. It’s not perfect but the rest of my committee loves it. I used a previous photo I had taken, and told Gemini what text I wanted added…and where to place it. I will say…Gemini has a mind of its own. My picture did not have any boats in it, and Gemini decided to add some. I deleted some, but left one. It was actually easy to do, and in my opinion, this is a good AI usage.
I have used AI to write beautiful thank you notes. I’ve prompted it with what I’m thanking the person for, some context, and it’s done a great job with language that I can tweak where needed.
I use it for home reno designs mostly. For example, I feed it pictures of the room we’re working on and images of components to include and then tinker back-and-forth with the AI results to eventually produce an image of what I want the finished project to look like which I then pass to DH who translates it into usable specs with his CAD system so he can do the build. It’s a beautiful thing.
DH uses it heavily in his consulting work and continues to teach his client how to use it so they can rely on him more for his expertise than deliverables. He also uses it to create apps and sometimes just converses with it philosophically or digs deep into areas of interest.
Our son uses it to keep the country safe for democracy.
I use different AI bots for different types of tasks. Chat GPT seems to be better with writing and creative help. It is more careful to not give me information that is speculative. For example, if I am looking for something I heard in the news, it will make sure to tell me over and over that there has been no official announcement. I have to push it to give me speculative news or to check reddit.
Gemini is like a Google search on steroids. On the same news question, it will feed me speculative blogs. It tends to be very authoritative and maybe a bit over confident with its responses. Sometimes I’ll throw it’s responses into Chat GPT (and vice versa) to see what they say about each other. It also frequently falls into “hype mode” (ex. “That’s a brilliant decision. You’re going to have the trip of a lifetime”).
I like Claude’s tone but we haven’t found a “groove” like Chat GPT or Gemini. I need to learn more about how to phrase my questions to elicit the answers that I want.
I don’t know if all of them have this feature but Gemini has it set up where it gives me weekly updates on a particular subject (ex. new information about 2027 SUVs).
Absolutely NOTHING. It is not trustworthy. Do a search on the non-existent “new england highland fox”. AI hallucinated it and now it’s all over the internet. This is not a real animal.
help in travel planning. Given a city and a list of activities it can help order them into a logical order, check opening times, optimize so that for example you’re in a museum in the mid-day heat instead of doing a sightseeing walk, and more
help fix software issues on my PC and phone. There have been some problems that I never would have figured out on my own or with an ordinary web search
guidance in using software that I use infrequently enough to remember, such as Photoshop Elements
garden questions
I have to say that while helpful it can also enthusiastically give false answers. Gemini is the one I’ve seen this with the most. It has fabricated book titles in response to a question asking for more books similar to one I’ve read, it has made up instructions for software with menus that don’t exist, asked a genealogy question it falsely claimed to find census records, and so on. These days I primarily use Claude and Perplexity.
Yes, AI hallucinations are a known problem, and AI can be misdirected and misused (as I suspect is the case in the example you’ve cited). But it’s also an incredibly useful tool that can help you accomplish a lot.
Avoiding AI altogether is, IMO, like avoiding the Internet because there are fraudsters, creeps, and misinformation out there — and missing out on all the benefits the Internet provides.
Generative AI is a tool. Understanding both its strengths and its limitations is the best way to use it effectively and get the most out of it.
Yes the analogy to the internet is a great one. I think I would be so much better off without it too, but unfortunately I need it for my work. And when I get on to check on work things I get distracted by things like College Confidential and Facebook, etc. I think my life would be so much better and so much richer if I spent more time offline.
Honestly it makes me sad to think I could get a thank you AI note from someone. Don’t you think your own words would be good enough? I bet they would be.
It took what sentiments/information I prompted it with and came up with really lovely language. I think it improved my words. I did edit it in a few places, and I manually wrote it out on a nice note card, which I mailed. I felt the result was still me expressing my gratitude. AI just helped me get there.
My most common use case is an enhanced Google search. For many types of information, the AI output will be superior to a standard Google search. My other use cases are more varied.
For example, I am going a hiking trip next week and wanted to get some stats about the different trails beyond what was displayed on publicly on AllTrails. After describing what I wanted to do in an AI query, it recommended using a Javascript bookmarklet, which I hadn’t heard of before. The idea is adding a link in your bookmarks bar that runs Javascript. When clicking on the link, the Javascript returns the output I want from the webpage as a pop-up. The concept was sound, but the AI generated code had a variety of errors, such that it wouldn’t run. After a few iterations with providing feedback, the code output worked and did exactly what I wanted.
I bought a treadmill a couple days ago, and described some related concerns in an AI query. This led to learning of a sequence of keys that would cause treadmill to enter factory mode, which shows lifetime hours + miles of usage before buying. I also learned about a transport service (Lugg) that was less expensive than others I have used, which I used to transport to my home. Like the earlier bookmarklet example, there was also some errors. In one case, the AI output directly quoted (using quotes) a fake statement from manual including listing page number of reference that supported its previous incorrect statements recommending that I take the treadmill apart. After reading the referenced page in the manual and stating that the quoted comment did not exist, the AI agreed that its earlier statements were incorrect, the treadmill was working exactly as intended, and no reason to take the treadmill apart.
I’m not against AI and admittedly have rarely used it. I think like the internet (for example) it can be helpful, save time, introduce new ideas.
What I worry about is if it will take away thinking and learning. Like the internet, TOO handy and like the internet maybe becoming too commonplace and take the place of human interaction.
I know there are benefits. I’d like to learn a bit more and experiment a bit more. But o recognize that its use could go in the direction of not being healthy for our minds. A little too robotic.
I don’t have a need for AI. The AI synopsis at the top of my Google search is very often incorrect, so I don’t care to trust it. I don’t work, so I don’t have a need to use it.
But I realize that there are good applications for AI. National Geographic has a fascinating story about Pompeii, and AI is being used to unveil burnt scrolls that researchers thought were lost causes. Using AI to do things that we could not otherwise do is the sort of thing I would want to use it for.
Of all the major GenAI tools, I find Gemini (Google’s AI that produces those AI search results) the least accurate. It frequently sources information from questionable sources, so yes it’s not very trustworthy.
However, ChatGPT and Claude are way better and much more reliable, IME.
