Theresa Payton | "Scraping" | A.I. shortcomings magnified in medical field

Theresa joins Bo and Beth to talk about Reddit blocking the internet archive from "A.I. scraping", as well as medical professionals raising concerns after an A.I. bot reportedly referenced body parts that didn't exist when attempting to diagnose a patient. 

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:00):
This is Good Morning Bet with both Hubson and Beth
trout Man, News Talk E eleven, ten, DO nine and three,
WBT BO and Beth on a Wednesday and on the
doublebt Hotline. Here she is once a week. We talked
to her and a lot of people across the country

(00:22):
do as well. You see her on CNBC, The Today Show,
Good Morning America, Fox and Friends. Teresa Payton, our cybersecurity expert,
founder of Forderless Solutions, Good morning to you.

Speaker 2 (00:34):
Good morning. It's great to be with you, and it's
great to be in Charlotte, which almost felt like it
wasn't gonna happen last night. I was delayed almost three hours.
Oh gosh, Chicago.

Speaker 1 (00:44):
Well if you like rain, If you like rain, welcome
to Charlotte.

Speaker 3 (00:47):
Yeah, a little soupy outside.

Speaker 1 (00:50):
So I have a lot of things to talk to
you about this week, but this is one to begin with.
And you know, we actually from time to time have
been known to use a Reddit list or two on
the show. Yes, we like Reddit. It's a very useful
at times. Reddit is blocking the Internet archive from AI scraping,
AI scraping. I know you just finished speaking at a

(01:12):
conference where AI came up left and right, and it
really does wherever you go, and understandably so, but what
is AI scraping and and why should we be concerned
with that as it relates to the Internet Archive.

Speaker 2 (01:25):
Sure, well, I mean so AI scraping. There's bots calling
the Internet all the time, and they were doing it
before we had chat GPT to look for information to
give people results, but also the collect information so just
almost like copy paste, think of, I don't know, automated plagiarism,
so to speak. But the thing about the Internet archive

(01:47):
is that is the wayback machine. It is the machine
that takes pictures of the Internet and stores it and
preserves it like a copy you sort of a historical record.
And Reddit said though they felt like the archives wayback
machine might have been also grabbing information not just for
historical purposes, but for training purposes, and so they are

(02:11):
not going to allow the wayback machine to crawl and
post detailed pages, comments and profiles. All they're going to
allow it to do is take a picture of what's
on the reddit dot com home page, So it'll just
be archiving insights, but not all the things that people
said about it. So very interesting move by Reddit.

Speaker 3 (02:31):
Well it sounds like it's a privacy move by Reddit,
just trying to protect the privacy of the people who
utilize those chat boards.

Speaker 2 (02:38):
No, you're right, bet, then that could be in it.
They could have had the whole Reddit user baseing. Wait
a minute, this doesn't seem fair. But again I would
tell everybody posters beware, if you put it out in
the digital universe, just assume delete doesn't mean delete unless
it was really important to you, and then it probably
will be lost in some computer craft and the internet.

(02:59):
Never forget.

Speaker 3 (03:01):
Well, I know that you said that you were in
Chicago for the last couple of days and AI came up.
But Chicago and the state of Illinois in the news
speaking of AI. Illinois has actually banned AI therapy as
some states begin to scrutinize chat bots. And you sent
us an article about these human AI relationships that are

(03:22):
just no longer science fiction. People are now using AI
in place of person to person contact, including things like
therapy sessions.

Speaker 2 (03:31):
Yeah, and again this is sort of we just don't
really have the regulations around this. So it's one thing,
if you're using AI, you have a doctor and they're
recommending we're going to use this program that we have
trained as a doctor's organization who's going to help you
with therapy so that you can have twenty four by

(03:52):
seven access. It's another thing when you create your own
chat personality and chatbot and then you're conversing with it,
and we're starting to see different articles come out. You know,
there was another there was a man looking for sort
of healthy advice and he removed salt from his diet
and was using some other alternative and he's actually very

(04:13):
sick right now. So we have to be very careful
here because just because it sounds authoritative, friendly, conversational, like
a person that's you know, your best friend always on demand,
doesn't mean that it's healthy and that it's good for us.
And that lack of regulation that we have in this
technology it's really creating a real problem.

Speaker 3 (04:34):
There was just an article that popped up about AI
making up body parts, trying to diagnose this was a
neurological diagnosis, but actually making up body parts. And it's
kind of along the lines of what you and I
have talked about before with AI coming up with a
way to lie just to sound like it has an answer,

(04:55):
and in this particular case, it was being used to
diagnose but made up a part of the that didn't exist.

Speaker 2 (05:02):
Yeah, this this one was This one was a really
rough one because this is actually something where Google was
creating a healthcare AI. But you know, we don't need
to pick on Google, like all of the different platforms
are having these issues and doctors were testing it out
and it was making up a body part that doesn't exist.
You're right, And I think that one of the things

(05:25):
people need to remember as they're interacting with tools like chat, GBT,
grock claud. You know, whatever your choice is that you're using,
is the way the engineers have created the technology. It
is supposed to give you an answer. And so in
trying to get to that, you know, the end of

(05:45):
the computer loop, so to speak, the computer program in
designing to give you an answer. It is not really
trained by most of the engineers to say, you know,
I'm sorry, I just don't know. Can you ask me
a different question? You know, back in the days, remember
when we would do Google searches. That sounds so quaint,
Beth and bo, but you would try to find something

(06:06):
and then it would be like error, you know, linked
not sound. You know, sometimes search gets it wrong and
we would see it so spectacularly when we get an
error message. Well, now it's as if, you know, these
jen Ai programs don't want to admit that they could
be wrong, and so they just make something up in
the end, just to give you an answer.

Speaker 1 (06:25):
Let me roll back just for a second before we
say goodbye. And you mentioned the Internet Archive, and I
brought it up, and then you said the Wayback Machine.
And I think we've touched on this once before, but
I think it's one of those things where somebody might
be driving along saying where exactly is the Internet Archive?
Is it in a room with padlocks on it at
the Smithsonian, you know, is it physically somewhere? And who

(06:47):
decides that is the Internet archive? You know, we've been
using the Internet since what nineteen ninety five ish, and
it's just so commonplace. But then where's the server for
the whole thing? And if somebody goes in and flips
the switch off, then we all go dark. I know
that's not how it is, but then again, how is it?

Speaker 2 (07:06):
Well? You have to wonder sometimes with sort of the
ups and downs of technology, you know, I often think
of that scene and airplane where he you know, unplugs
the h airport, but it is. If anybody's interested in
finding the Wayback Machine, also known as the Internet Archive,
you can go to web dot archive dot org and

(07:27):
you will find the nonprofit there.

Speaker 1 (07:30):
So there it is. It actually is more of a
physical thing than I and we thought, and we thought.
So that's why we love talking to Teresa. The the
incredibly complex questions and then questions like where is the internet?
And you know, how long will it take me to
drive there? All right, it's good to talk to.

Speaker 2 (07:46):
You well, Beth and Bow. By the way, the Wayback
Machine has over nine hundred and forty billion web pages
saved over time, so it take you a while to
get through it. But Beth and Bow, it's always great
to be with you and be safe out there.