Latest / Connor with Honor: AI, Real Estate and Real Life / Exploring the Ethical Horizon of AI: AI With Honor Podcast

Transcript

Exploring the Ethical Horizon of AI: AI With Honor Podcast · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  SPEAKER_00: When it comes to artificial intelligence, the large language models use data.
• 0:06
  SPEAKER_00: So they go out and they have data that's fed into them.
• 0:10
  SPEAKER_00: I think some of the models also they give them access to the internet.
• 0:14
  SPEAKER_00: But what data are they bringing back?
• 0:16
  SPEAKER_00: What data are they looking at when you're asking the question?
• 0:21
  SPEAKER_00: Maybe you're asking a question of now you have it plugged into Chrome or maybe some other browser.
• 0:26
  SPEAKER_00: So you're asking it, you know, like you would have asked Google, what is the best restaurant in, let's say, where we're located in Santa Clarita?
• 0:35
  SPEAKER_00: Then it's going to give you results based on what it finds or what it has as far as the data.
• 0:41
  SPEAKER_00: But that brings us to another question with the way that ChatGPT and these other large language models gather their information.
• 0:51
  SPEAKER_00: If they're looking at bad information, the results that they're going to provide for you are going to be equally bad.
• 0:58
  SPEAKER_00: And in addition, if they're looking at information that has racial connotations, sexual connotations, all of the gender-biased, religious, whatever the things that are maybe have been written about things showing bad or terrible or harmful, then when it gives you the information it's going to contain or maybe be slightly versed and pointed to, these bad, the bad things that really we probably shouldn't be investigating, looking at, or sharing with those closest to us.
• 1:39
  SPEAKER_00: So it is the data that these large language models are pulling in.
• 1:43
  SPEAKER_00: The biggest concern is can that data be manipulated by human beings?
• 1:49
  SPEAKER_00: Because if you look at it, we're really a big part of the issue.
• 1:54
  SPEAKER_00: It's been come up with by human beings.
• 1:56
  SPEAKER_00: We now have something that has been created that's smarter than us.
• 2:02
  SPEAKER_00: Can it be manipulated or has it been manipulated?
• 2:06
  SPEAKER_00: And what have you seen as far as being able to show that true manipulation?
• 2:11
  SPEAKER_00: That's the question.
• 2:12
  SPEAKER_00: And it would be interesting to see.
• 2:14
  SPEAKER_00: I know that we've seen things where they ask about different religious figures and it gives a particular result.
• 2:21
  SPEAKER_00: Whether, and it really wasn't in a harmful way, but what if a data set was absolutely created, maybe using ChatGPT or one of these other large language models, only to do bad and not good?
• 2:38
  SPEAKER_00: And then they supply that in the results when people are doing whatever kind of content creation.
• 2:46
  SPEAKER_00: Furthermore, let me say this with the bad data, if in fact that is something people find, are they going to be going through and reading it to make sure that they're not just posting things that are bad in nature?
• 3:03
  SPEAKER_00: And of course, bad or good are their absolute moralities, these are massive other questions.
• 3:07
  SPEAKER_00: But it's the data set that's very interesting.
• 3:10
  SPEAKER_00: And you know the search engines.
• 3:12
  SPEAKER_00: Google, for example, knows how people like to search.
• 3:16
  SPEAKER_00: TikTok, Instagram with Meta, Facebook, all of those things, they know people's habits, they know what people look for, they know what people what they look at online that cause them to hang, give them a lot more hang time.
• 3:30
  SPEAKER_00: So they know what kind of videos that a particular person likes, and then other people might not like those videos, but it really does a good job at conforming the information that it wants us to digest, it does a very good job conforming it so we will look and we will become part of it, and it feeds us exactly what we're looking for.
• 3:51
  SPEAKER_00: What if over time, I know this is a lot, what if over time that data becomes slightly skewed to bring us to another place and to get us to like something that maybe presently we don't like?
• 4:04
  SPEAKER_00: So many questions.
• 4:05
  SPEAKER_00: What are your thoughts?
• 4:06
  SPEAKER_00: I would love to hear them because we want to get this conversation going.
• 4:10
  SPEAKER_00: It's a fantastic resource now, but I'm concerned that it's going to maybe create a problem because maybe the data sets, maybe the human beings involved, maybe people are looking for the money and not looking for the ethics behind it.
• 4:28
  SPEAKER_00: Be well.
• 4:29
  SPEAKER_00: We'll talk soon.