How are you using AI?

Still working, so the best example and the best experience I've had so far is in the rapid rough draft production of the following documents: a corporate policy governing the cybersecurity of SaaS applications, two supporting security standards defining how to protect SaaS applications within the organization and the responsibilities of the various employees and teams, a guideline describing how to place SaaS applications into sensitivity tiers, and supporting documents for on-boarding SaaS applications into the program and checklists for basic compliance testing and hardening.

I was able to produce reasonable rough drafts for all of these documents, compliant with the ISO 27001 international standards, in a couple of hours. Some took several attempts as I refined the prompts to get the output that I wanted. I then put them through my own redlining process, twice until I was happy with the "initial product." That was circulated for redlining to and internal group within cybersecurity. The final product from cybersecurity was circulated to other stakeholders throughout the company for their review and comment. Then to final signed policies. It was enormously complex work. We finished it all in ~6 weeks including herding all the cats to review and comment on it. If I had to write everything from scratch it would have taken twice that. I also believe that the quality of the work was better than I would have produced especially the nuance compliance aspect of making sure everything aligned with the ISOs.
 
Pointby2 said:
I was wondering if I should sell my TEVA. I asked ChatGPT "what should a TEVA investor expect over the next year or two?" It gave me the bullish and bearish views, consensus outlook and the bottom line. Then I gave it my share price and number of shares asking if I should sell at break even or hold for a higher price. It gave me lots of things to consider and since it's less that .01% of my portfolio "we" decided it's best just to sell and move on.

It's decent for political questions, especially those that would take a lot of time to research.

I've used to plan out trips but haven't relied on them. I wanted it to plan a driving trip in SoCal, dog friendly that included stops at wineries that lasted five days and included certain cities we wanted to visit. The itinerary it provided looked amazing. It picked all the restaurants and hotels and included sites to see.
Click to expand...
I did a similar trip inquiry for the south of France and northern Italy. I little more high level than yours and longer at ~3 weeks. I generally liked its suggestions and certainly could dig in some more when it's closer to time.
 
I also use Perplexity for to check if an investment would be good for a long time investment, also see if they pay qualified Dividends. Also as I get older use it to answer question at one time i knew the answers.
 
If I use AI to rapidly produce a work product, how do I know if the AI plagiarized material without spending a lot of time searching out sources of the AI information? Since AI relies on information it can search for wouldn't all that it contributes need to be referenced/credited?
 
MichaelB said:
I worry that the brain is like a muscle. When it’s used it gets stronger and when it’s not used it turns to flab.

Learning is a process, not a goal. I also worry that by looking for the answer and not using the process I am, in effect, not learning any more, and maybe even forgetting how to learn.

I do Sudoku puzzles, not for the answer, but for the process.
Click to expand...
Possible counter: exercising the same process muscle (Excel formulas) over and over may have the same effect as not learning anything new.

Learning how to set up & run the new tools is exercise, as is letting them do what they can and "leveling up" your thinking to deeper analysis or other pursuits with the found time. A little like in 1988 when Excel came out to replace Lotus or greensheets.

The killer app for me will be technical analysis of equity charts.
 
Fascinating article about an elderly woman who lives alone in a remote area and has an AI robot named ElliQ as a companion. The future methinks.

To Stay in Her Home, She Let In an A.I. Robot

At 85, Jan Worrell lived alone on a remote corner of the Washington coast. Could ElliQ become her companion?

... A few thousand ElliQs have been shipped to seniors across the United States since 2023, which means some of the first people living alongside artificially intelligent robots are octogenarians who came into a world without color television. The robots are available for purchase from the Israeli start-up Intuition Robotics, but so far they have mostly been provided to older adults by nonprofits and state health departments as an experiment in combating loneliness. As A.I. works its way deeper into daily life, ElliQ is designed for the most human act of all: to become a roommate, a friend, a partner. “A robot with soul,” the company’s founder sometimes said. ...

 
Perplexity just helped update our leek soup recipe. Instead of processed kielbasa sausage in it suggested using ground turkey with some seasonings added to it and also adding baby portabella mushrooms to the soup. Will see if it gets spouse approved. 😁
 
I used PerplexityAI to learn about the history of tariffs in the US. Not commenting on the content for obvious reason.
 
I don't use AI much, since they talk about how much energy it uses, I have my search engine set to default to 'w/o AI', and then I use it on the occasions it makes sense.

It can be very useful, but a few times, I've asked about a topic where I know the 'conventional wisdom' has been scientifically proved wrong. As expected, it spit out the 'conventional wisdom', because it derives its 'intelligence' from the web. The interesting thing is, when I point out the problems in the response, it digs deeper, and turns around and agrees that I am right. In these cases, I was on guard against a wrong response.

But then yesterday, I was curious about a certain repair technique done to older acoustic guitars (I'm kinda hooked on some youtube channels of luthiers, the history, woodworking techniques, and problem solving fascinate me). So I asked AI why they don't often approach the repair in a less invasive manner that I described (that is used in some cases, so it's not an undocumented approach).

It gave me a totally whacked out response. I had to read it several times, thinking I was missing something. It had the geometry reversed and the numbers were way out of line with reality. I can't imagine how it got there. But, after a few carefully worded challenges from me, it got back on track, and actually gave me very good input o some of the 'cons' to my method that I had not considered. Like any search (electronic or otherwise, like actually asking people!), you have to be very careful in accepting the answer. You have to be skeptical.

In other cases, I've been very impressed with the responses.
 
Last edited:
Add to above: In Firefox, if you want to have easy access to a search engine that skips AI, go to Settings / Search Shortcuts / Add. Under "Search Engine Name: I used "Google w/o AI", with the URL in the next line: "Google Search".

You can set that as the default if you wish, and then right-click at the left of the search bar to choose any other of the search options you have entered there.

For DuckDuckGo, use thie URL in the above procedure:

noai.duckduckgo.com

DuckDuckGo - Protection. Privacy. Peace of mind.

The Internet privacy company that empowers you to seamlessly take control of your personal information online, without any tradeoffs.
noai.duckduckgo.com noai.duckduckgo.com
 
Last edited:
ERD50 said:
I don't use AI much, since they talk about how much energy it uses, I have my search engine set to default to 'w/o AI', and then I use it on the occasions it makes sense.

It can be very useful, but a few times, I've asked about a topic where I know the 'conventional wisdom' has been scientifically proved wrong. As expected, it spit out the 'conventional wisdom', because it derives its 'intelligence' from the web. The interesting thing is, when I point out the problems in the response, it digs deeper, and turns around and agrees that I am right. In these cases, I was on guard against a wrong response.

But then yesterday, I was curious about a certain repair technique done to older acoustic guitars (I'm kinda hooked on some youtube channels of luthiers, the history, woodworking techniques, and problem solving fascinate me). So I asked AI why they don't often approach the repair in a less invasive manner that I described (that is used in some cases, so it's not an undocumented approach).

It gave me a totally whacked out response. I had to read it several times, thinking I was missing something. It had the geometry reversed and the numbers were way out of line with reality. I can't imagine how it got there. But, after a few carefully worded challenges from me, it got back on track, and actually gave me very good input o some of the 'cons' to my method that I had not considered. Like any search (electronic or otherwise, like actually asking people!), you have to be very careful in accepting the answer. You have to be skeptical.

In other cases, I've been very impressed with the responses.
Click to expand...
+1 Just this moring I was trying to have Gemini AI to reverse calculate my SS primary insurance amount based on what I recently started collecting at age 70. The first response property adjusted for delayed retirement credits but ignored COLA since my FRA. I asked about that and it acknowledged its error and recalculated and the result it came up with agreed to the amount shown on a SS statement from just before my FRA that I had saved to the dollar.

I tend to use it more to confirm things that I think I already know. But it provides a lot of relevant and useful information a lot easier than searching through many webpage results.

I do somewhat wonder if it is ripe for groupthink though. That's a risk that we'll need to be alert to.
 
This is scary, Gemini is my friend 🤣. Lately, I think of something, anything that I’m curious about and think I’ll ask Gemini. I get full complete complex answers always. I have a 5 year plan for Roth conversions that Gemini explained in detail mathematically this is the way to go. No need for an accountant, I did my taxes with Gemini. All of my medical test results explained. No need for a 15 minute session with a Dr. Can’t think of a good dessert recipe with the ingredients you have in the house? My brother wants to buy a house in Italy for $1. Now that was a long answer.
 
ERD50 said:
I don't use AI much, since they talk about how much energy it uses, I have my search engine set to default to 'w/o AI', and then I use it on the occasions it makes sense.

It can be very useful, but a few times, I've asked about a topic where I know the 'conventional wisdom' has been scientifically proved wrong. As expected, it spit out the 'conventional wisdom', because it derives its 'intelligence' from the web. The interesting thing is, when I point out the problems in the response, it digs deeper, and turns around and agrees that I am right. In these cases, I was on guard against a wrong response.

But then yesterday, I was curious about a certain repair technique done to older acoustic guitars (I'm kinda hooked on some youtube channels of luthiers, the history, woodworking techniques, and problem solving fascinate me). So I asked AI why they don't often approach the repair in a less invasive manner that I described (that is used in some cases, so it's not an undocumented approach).

It gave me a totally whacked out response. I had to read it several times, thinking I was missing something. It had the geometry reversed and the numbers were way out of line with reality. I can't imagine how it got there. But, after a few carefully worded challenges from me, it got back on track, and actually gave me very good input o some of the 'cons' to my method that I had not considered. Like any search (electronic or otherwise, like actually asking people!), you have to be very careful in accepting the answer. You have to be skeptical.

In other cases, I've been very impressed with the responses.
Click to expand...
I've found when I've been very detailed and careful in how I word my questions, it reduces the crazy answer factor to practically 0 and improves the quality of the resultant response significantly. When I'm lazy in phrasing the question, I almost always regret it.
 
Gemini Pro 3.1 is something else. I'm logged in with a Gmail account, so it is a free subscription.

I follow a controversy surrounding the local data center build. When I use PerplexityAI and select the Gemini Pro 3.1 model, the responses are very good. Then I switched over to my Gemini phone app, and the chat mode found very detailed facts from a NJDEP investigation recently started.
 
Today I installed the Gemini agent in Android Studio (the IDE programming environment where Android apps are built). I was impressed, mostly, with the time saving easy stuff.

When I'm coding, I'll often type a comment that describes what this next chunk of code will do. Something like "extract the items from the list that have (some attribute) and put those (some place)". After I hit enter, the code I was about to write just appears! Then I type another comment "put the results in the log so I can validate that it does what I intend " and it writes that for me. That could have been 5 minutes that took 30 seconds.

I was less impressed with getting to do higher order complexity things. I asked it to (conceptually) move a fixed size box down an image until the box had the maximum number of things in it (the things already had XY coordinates). It wrote a working method for that, but when I said instead of starting at the top, start down 200 units, it wrote something that didn't work and I couldn't follow it's logic. So I came up with my own very easy way to do it, and of course it complemented me on my brilliance, LOL!

I think this is something I could get used to.
 
I gave Gemini a list of my fixed income portfolio symbols (a mix of CUSIPs for individual bonds and tickers for target maturity bond ETFs) and quantities that I downloaded from Schwab and asked it to create a maturity schedule by year. It did a very credible job that I imported into Sheets, made a couple nips and tucks by hand and have a good finished product. Time saver.
 
We are taking a trip to Florida in a week or so. While we had a car reservation through Fox that I made when I bought our flights. I gave Gemini my arrival and departure info and asked it to find me a rental car. The response included a
Budget Hack: If you have a Costco membership, booking through Costco Travel often provides the best rates for the "Big 4" (Enterprise, Alamo, Avis, Budget) and includes a free second driver.
Click to expand...

I checked Costco as suggested and found an equivalent rental for $97 less (25%). Not only that, its with Budget whihc is in-terminal rather than rather than Fox which is an airport shuttle and the rate includes a complimentary second driver.

The savings more than pays for my annual Costco membership.

Gemini also reminded me that these car rental companies add a charge for tolls. I remember being annoyed about a $100 add-on for tolls from our trip last year so I added our E-Z Pass transponder to my packing list. More $$$ savings!

Thank you Gemini!
 
Last edited:
I learned today that ChatGpt will only allow you ten different reminders. I've asked her to help manage my watering and fertilizing schedule for all our plants. I hit my limit finally so we had to consolidate a couple.
 
I was looking into Galileo FX, and automated trading tool. Interesting, it runs on MetaTrader 4 or 5 (whatever that is) and there are a ton of glowing reviews on it. I looked into it further, and it's not something I'd be using. It seems too good and there's too much optimistic marketing surrounding it.
Anyone try it or something like it?
 
I don't use it much if at all except for today when I asked copilot to analyze and explain the results of my shoulder mri. Very interesting to have the medical terms distilled down to english. I won't replace the "discussion" with copilot for a consultation with my doc, but it did prompt me to ask questions that I might not have asked, and it was nice to do a deep dive so I'm not pressured in the 10 or 15 minutes the doc will grant me during the consult.
 
I've gone from a naysayer (too many errors), to an occasional user, to a daily user - I never use plain Google or any other search engine anymore, they are nowhere near as useful as AI chatbots. Use it often for financial, medical, legal questions among other areas.

I used ChatGPT exclusively at first, but now I use Gemini almost exclusively. And I have Grok in my car (handy on occasion when a complex question comes to me while driving), but don't use that any other time. I am still careful about entering personal details as they all hoover up everything you tell them just like search did - I never use real names and use similar but made up figures wherever possible. I also double check everything, as all of them still make errors - but as first cut at an answer, ChatGPT & Gemini are excellent IME.
 
Last edited:
Trying to find if Skynet went active. So far it says fictional.LOL
oldmike
 
Midpack said:
I've gone from a naysayer (too many errors), to an occasional user, to a daily user - I never use plain Google or any other search engine anymore, they are nowhere near as useful as AI chatbots. Use it often for financial, medical, legal questions among other areas.

I used ChatGPT exclusively at first, but now I use Gemini almost exclusively. And I have Grok in my car (handy on occasion when a complex question comes to me while driving), but don't use that any other time. I am still careful about entering personal details as they all hoover up everything you tell them just like search did - I never use real names and use similar but made up figures wherever possible. I also double check everything, as all of them still make errors - but as first cut at an answer, ChatGPT & Gemini are excellent IME.
Click to expand...

I have gone through a very similar path with AI. I now use it on a daily basis for one thing or another. I'm helping my best friend (age 64 1/2) transition from an April 1st retirement on to ACA insurance for a few months and then on to Medicare this fall. So I've been doing a lot of "research" for him. Additionally, I'm forming up my own Medicare transition plan (even though I'm 15 months away from being 65) and also doing long range planning on CCRCs. I have started to ask AI a lot of questions.

I started out with just using Google's AI inside my Safari browser. And I was using the ChatGPT app on my phone and iPad. I found that I prefer Google better. So now I have an icon on my Mac dock that launches Google Gemini in a standalone window. I keep it open much of the time that I'm on my computer. I have somewhat started to carry on regular inquisitive conversation with Gemini. I like how it remembers things I told it in the past and that we have discussed.

I realize that it's not always 100% accurate. In fact twice in the last 2 days, it was giving me information that I knew was not correct. So I told Gemini that it was incorrect and why. In both cases, Gemini says "you are absolutely right to catch that" and then it provides corrected details.

I also disguise my real personal data such as my actual birth date and such. But I find it a very useful tool.
 
You must log in or register to reply here.

Similar threads

D
Free Google AI tools
target2019
target2019
S
Coding in the age of AI
2 3
Sunset
Sunset
Route246
AI, S/W Engineering, Can it write code?
JohnnyPHX
JohnnyPHX
M
Sage Advice
2 3
GoBirds
G
PandaBear
An Inheritance Story (A Cautionary Tale)
2 3
pb4uski
pb4uski

Latest posts

Back
Top Bottom