Expert reveals the five things you should never use AI for

Fox - 2 Detroit
22 followers
831 Video Views·Sep 5, 2026
AI Generated·May contain errors
Well, three hikers needed to be rescued from Mount Shasta
after admitting that they relied too heavily
on AI to plan their trip.
They did not have enough food, water, or emergency gear,
and eventually got lost on the way down.
It's a pretty extreme example,
but it is a reminder that AI can sound confident
while getting things wrong.
And tech expert Joe Tavares joins us again now
this Friday night to talk about what went wrong
for these hikers and five things
you should never rely on AI to do.
So Joe, this is a pretty wild example,
but what do you make of this thing here?
Well, folks are getting a little bit more comfortable
with AI and the advice that they're getting from it.
And these folks asked what the best route is
to take up Mount Shasta, got lost, didn't have enough food,
and realized that the map that they had wasn't accurate
to the trail they were trying to hike.
So this is one example, I guess,
of five things that we should never ask AI to do.
What are some of those examples?
Yeah, so number one, obviously,
don't ask it to be your Sherpa.
Don't try to have it guide you up a mountain.
Number two, I would say,
don't ask it specific health details.
It certainly knows quite a few things about medications
and maybe contraindications, things like that.
But AI can make mistakes,
and I don't think anybody wants to go to the doctor
and have the doctor tell them,
hey, I can make mistakes sometimes.
So that seems like one of the ones
that you should definitely rely on professionals for.
Number three is probably lawyers, legal, law.
There's been several cases now of hallucinations,
making up case law, courts throwing out cases,
things like that, because lawyers are using the AI,
and if lawyers can't do it,
the average person's probably not gonna do too well.
Number four, psychological help.
So if you're having a mental health crisis,
there's phone numbers you can call.
There are professionals that do this for a living.
AI is reflective.
If you're having a rough time,
AI is not going to help you out of that.
It's going to reflect what you're saying
and drive you potentially down a darker path.
And last but not least, they're not good at math.
So I wouldn't trust a large language model
to do my accounting.
Once again, all these are really professional fields.
AI is prone to making mistakes,
and you should trust professionals for those things.
I have noticed, we call it Dr. Google in the old days,
when people would go onto Google to see about,
I have these symptoms, do I have this or do I have that?
So people doing that on AI,
it can take you down a whole different rabbit hole.
Absolutely, because when you look at the Mayo Clinic,
they have a specific article that's highlighting symptoms
of a specific disease or malignment.
And when AI is doing it,
it's just looking across the entire surface
of that physical malady
and perhaps pulling in information that's not relevant,
perhaps mixing up things like different medications
that shouldn't be used together.
Again, it's a broad search.
What is AI best used for in everyday life?
I guess for adults.
I find it good for general advice,
like, hey, it's 80 degrees, should I wear a jacket?
Might be a good one.
Things that are inconsequential
where you're not gonna get lost on a mountain,
I would say is probably the best use.
And maybe don't use AI if you have a restaurant
to make up what could be on the menu.
There's a national story about some really gross images
that on a menu that it's like,
okay, that's not what the restaurant really serves,
but yeah, so AI slap, avoid that, right?
Yeah, exactly.
And especially in the professional settings.
Judge Tavares, always a pleasure.
Thank you so much for coming on.
Woody.
