With AI, Always "Do Not Trust And Always Verify"

I did a quick survey of people who uses AI consistently to see what they are using it for. A vast majority of people I know are using it instead of regular Google search. Judging by Google’s earnings, I do not think the company is too concerned by AI taking shares away from them just yet. Having said that, what is concerning is that a lot of the people who do use search via ChatGPT, Gemini, or another LLM take what they are presented as fact. I have a different view of search via AI: never trust and always verify. 

For me, posing questions questions to the AI bots is a starting point that allows me to thinking deeper and eventually search the traditional way through DuckDuckGo (and then Google if I cannot find what I want right away). The people who uses AI as a search engine falls into two camps - those who just trust what the AI presents and those have a search flow similar to mine. A few months ago, I had an interesting use during a drive with ChatGPT’s voice function. We chatted with the bot about different types of trusts and how to set them up. It was a good 30 minute conversation that sort of set us on the path towards understanding how different types of trusts work. It was too much information at first so we told ChatGPT what we want and it provided information about it. I switched to Gemini to confirm what I learned and received similar information. This helped us gain a basic understanding so when the time comes to speak with a lawyer, we would have some sort of idea of where to start and not have to spend the time and money having the lawyer explain everything from the beginning. 

Using AI chatbots is a good starting place. It saves a lot of time and possibly money as in the example above. What’s important to keep in mind is that your leg work has not started. I would not consider a chat with the AI with something as important as the financial future of your family even the first step. Treat it as a start point. You still need more research to confirm what you learned and to talk to the right expert about the matter or subject. It is the same if you wanted to talk to a CPA, teacher or professor, or even a physician. 

Then there are casual AI chatbot interactions. I've asked about the Punch, the Japanese baby monkey, and how he is doing on a daily basis. I've asked about the current events in Ukraine, Iran, and East Asia. There are a lot of how-to's that I found very help, more than what I can find from the search engines. We recently installed a solar system with a battery and both Gemini and ChatGPT were helpful about scheduling appliance uses including the AC and when to charge our EV. All these do not need any verification. If one of the AI chatbots suggests that I can use solar power to power my house at night, I think I can catch on quickly that some wires got crossed in the LLM's reasoning. 

As always, use your best judgment. And think carefully about how to use AI and what the information the AI provides you. 

Comments

Popular posts from this blog

Created An App And Other AI Uses This Week

Google Contact Lenses, Apple’s Vision Pro, And The Future of AR in Glasses

Signing Into iCloud On iPhone Helps Get Around One iCloud Account Per Device Limitation