AI Literacy: The Key to Responsible Use of AI in Education | Mary Lou Maher | TEDxUNCCharlotte

do you remember the first time you heard about AI artificial intelligence I asked because your first impression isimportant sticky maybe you heard about AI by watching a movie uh 2001 Space Odyssey um how remember now the artificial intelligence on the spaceship well things didn't go well for the humans in that movie because the AI put its interests and its goals above what the human thought should happen so if you that's your first impression you may be carrying along this concern that AI is going to take over humans are going to become irrelevant or maybe watch

the movie her and in that movie AI and a man and fell in love it was an AI operating system and if that's your first impression maybe you think well maybe AI is going to replace our human partners because the AI can always be caring and loving and is never mean or nasty what will happen to us in our human relationships or maybe like me you learned about AI by being asked to study it as part of your PhD Research Now what's interesting about that is I'm an optimist I think we're developing AI for good

things we're making AI smarter we're making AI better now all of those first impressions stick with you the problem with your first impressions is it takes a long time to change them so what we what I want to talk talk about today is what do we do about that because there's a lot of AI out there and there's a lot of variation in these first impressions and what I'm proposing is that AI literacy is really important and very essential that AI literacy in fact will help us gain control over the AI That's out there it

will help us gain control of the narrative to be more aware to be more critical and more understanding of what we're dealing with so what is AI literacy well AI literacy is a set of competencies firstly it's understanding the definition and the scope of AI now many people think that AI just started it was in the movies sure a few years ago but it wasn't real but there actually were real AI systems out there as early as the 50s so it's not that new understanding that traj trory is really important you also need to understand

the ethical issues that are associated with AI and they come from two different directions the ethical issues of the people who create the AI and the biases they bring to that but also the ethical issues in the use of AI and then finally AI literacy is a competency associated with the responsible use of AI most of us are not going to build our own AI systems we are going to use systems built by others how do we do that responsibly and what this does is it develops a set of competencies someone who has AI literacy

is actually able to think critically about AI Technologies someone with AI literacy is actually able to communicate and collaborate responsibly and effectively with AI and to use AI responsibly so why do we need AI literacy now why am I up here talking to you about it now why not 20 years ago 10 years ago 5 years ago it's been around why is it an imperative now that we have ai literacy well firstly the widespread use public use uh it's very interesting to reflect on that uh open AI released chat GPT in November 2022 but by

January 2023 two to three months later it had 100 million users that's amazing and all that happened in that two months when many of those 100 million users did not know what they were interacting with the second urgency is that chat GPT this latest release of AI is kind of different from the other AI tools we've used first of all it's conversational now we have had chatbots before and the conversation around chatbots is how frustrating they are you know that they didn't understand my question uh or they answered the wrong question or they didn't have

an answer at all but chat GPT always has an answer and chat GPT is surprisingly competent at understanding your question so it's a different kind of conversation it's much more human it doesn't feel like a chat bot it feels like you're actually talking to a person the other reason chat GPT is different from those chat Bots or other AI systems is that it's very has a very broad general knowledge it's been trained on very large datab basis and because of that it can respond to almost anything you ask it's not trained on a very specialized

area and then finally the concerns you know this morning I had to go pick up my car and the people were helping me with the car were just chatting with me and they said so what are you doing today and I said oh I'm going to give a TED Talk and they said oh a TED Talk that's exciting what are you talking about and I said Ai and they said well that's scary so everyone has these first impressions the scary part is something that is being called in the popular press hallucinations hallucinations are when you

ask chat GPT to do something and it tells you something that's incorrect if you don't know why that can happen happen then you need to become AI literate it's not that it made a mistake it's based on a prediction system that was the most likely answer to that question based on its model why can it be wrong privacy issues you know you can have an interaction with chat GPT but if you're on the public version which is where most people are then everything that you type into chat GPT is part of its next round of

training so there's no privacy so if you're a student and you put your paper in to the prompt and you say make this paper sound more professional that paper is now available for its next training session and then finally the inappropriate use all good technology can be used for bad and without AI literacy we don't even know how to recognize that so we're teaching a course on AI literacy here at UNCC Charlotte and it has these three parts first understanding the scope of AI understanding both the history how we got to where we are now

what's the difference between AI based on deep learning models and AI based on cognition and how people reason chat GPT is not based on how people reason it's based on data we go in a little bit into how does it actually work how does that prediction happen when you interact with chat GPT but this is an AI literacy course it isn't an AI course in an AI course you'll learn about the mathematical models that underly the algorithms you'll learn about the probabilistic reasoning you'll be asked to build a deep learning model in this course it's

about interacting with using and understanding how it works enough so that you can use it critically we also talk about the difference between something like chat GPT and say Google search or Bing search for many students they they didn't actually appreciate the difference till the end of the semester but if you don't understand that difference then you don't understand why chat GPT can tell you something wrong where being will give you a list of web pages to look at chat GPT is very conversational it's very friendly very eager to please you they're completely different behind

the scenes and understanding that difference again helps you understand should I do a search in Google or should I use chat GPT for this we did focus more on how large language models work because that's what chat GPT is but we went beyond that we talked about the ethical and the responsible use of chat GPT that isn't as simple as knowing that there's biases in the data or that there's unethical use of it but how do you actually prompt it we're used to putting a prompt of sorts into a search engine but a prompt into

a chat GPT need not be something that says tell me the answer to this you put your homework assignment in there it'll take that many characters and then it gives you the answer to the homework assignment well that's not that responsible actually because you didn't do any of the learning there are ways to prompt chat GPT so that you prompt it to help you learn so you prompt it to ask you questions about a certain topic you're going to have a test on you can prompt it to do a a practice test with you and

tell you when you get the answers right and wrong so we spent quite a lot of time thinking about how do you prompt how do you interact with this new kind of AI and then finally we talked about societal implications because it's not just about us or me it's actually about the society we live in we talked about Ai and government policy Biden's executive order on AI the AI Bill of Rights we talked about Ai and the future of work when these students graduate AI will be part of their work environment what does that mean

for them and what kind of work that they'll do and finally we worried about public perception of AI because as I said first impressions are very sticky and you you may be talking to someone about how you took a course on AI and they go oh my God AI That's so scary student Reflections were really interesting it was a weekly exercise that we did in the course to find out what it was that they learned or thought they learned reflection on and what it was that they wanted to learn more about for example one student

wanted to learn how to prompt the AI so that they wouldn't bypass the learning so they wouldn't become too dependent on the AI another person said they wanted to learn more about this idea of who owns the information that chat GPT gives back can the AI actually own that material they thought AI was just a tool can it have ownership so where are we and what's next for AI literacy well one of the reasons for being here is to convince you that we all have to advocate for this both for ourselves for our friends for

the public we need to be AI literate we need to Advocate it for it as a course for all students not just University students all students need to know what AI is and then it's not enough for students stents to know in fact that would be a problem wouldn't it if only the students knew how to use AI we need to advocate for AI Literacy for all faculty and for staff as they do their jobs how can they interact with AI in a way that makes their jobs and their work better instead of in a

way that may one day replace them and then finally we need to Advocate to remove those barriers to access to AI at un iers ities and schools generally because the free AI is not the AI that's going to be the best version for our students it's going to be the one that now costs money and if there isn't Equitable access then we create a digital divide only this time it's over AI thank [Music] you