AI: Making Us Efficient, Lazy… or Just Efficiently Lazy?

Hope you're enjoying the conference so far. some great talks and I think what better way to follow you know we talked about attacks let's talk about impact how the AI is impacting us what exactly is happening what's the fun or what's the buzz behind AI and how it's relevant to us let's begin this is a background by me I'm a cyber incident responder at uh McKessen did my PhD from UAB and led investigations disrupted some malware and some previous works that I have done. So what is AI? Basically a computer system that can mimic human cognitive function and make decisions based on the data provided for training. Seems simple, right? And AI is everywhere, I'm pretty sure. So just just a quick

pulse like what do you think since morning? How many times have you interacted with AI? anyone five 10 20 five times uh probably more than that I'm sure you have social media accounts every news feed that you get is AI generated you have checked the weather app AI is using it if you have traveled here use GPS AI is there anything if you have done work if you have written an email if you have uh summarized any of the talks if you have the chart GPD on right now recording me you are using AI there is no getting away you don't even realize that it's just you know every ad that you see anything that you can think of

you have AI we are deeply involved with AI and we just don't realize it so this was a case study that happened uh some time back between Harvard uh university and open AAI they analyze 1.5 million chat had GPT conversations and uh the highest what it was found was that AI is used for writing and research that's the main I I was shocked with this the reason it it all was like yes the AI is going to be used for better coding doing those type of stuff but it's only 4%. Everyone is using for writing, researching and therapy 2 to 3%. People are using that and we're going to discuss that's that's very dangerous. AI

is not your therapist. I can tell you that and it's more like personal work than the uh use for work. It's like personal use that they are doing and women and young users constitute more than 50%. That was a surprising feature too. But yes, it is quite common in the young people. Everybody they are hopping on to it. This is our next generation. I mean yes we are old but the 18 to 25 they are coming in and they are using AI left, right and center and the growth is enormous. Organizations and society face structural shifts. Everything is moving towards AI and there is high dependency and human decision-m ability is being reduced. If you have AI to do your work,

you really don't want to put pressure on your brain or do the the critical thinking piece is gone and that's causing the problem. We have 900 million users worldwide. So I mean let's talk about the personal impact. How many of you here think that AI has made us efficient? What about lazy? And who all think like efficiently lazy? >> Yeah. So, so it is a combination I would say and we'll we'll do the this whole poll again at the end of the conference at the end of the presentation and we'll see if there is some change in that or not. You know efficiency let's let's take some examples of what efficient looks like. It can summarize your

documents. It can write summaries for you. It is as your assistant. It can read documents. you give it like nobody wants to do a research paper of 10 pages and read like 50 articles about it and get a topic and make introduction and whatnot. AI does it for you. It's it's helpful in that way that it summarizes it gives you the bullet points give you the gist of it. And this is a good area like cyber incident response. Of course, it's my field. That's why I really like it. Like it's helps reduce the false positives or the noisy alerts. It helps in threat intelligence and detection. There is so much out there that it is

difficult for an analyst to actually research everything. So that's where it helps. It helps getting like vulnerabilities, errors that are not previously seen. And this last point is my favorite. How many of you must I'm pretty sure every one of you must have hundreds of ideas that you want to work on. It's you never took the initiative for it. It's like okay I don't know I need to learn let's assume programming. I need to learn Python. Oh whom should I ask? I feel embarrassed you know at this age will I go and ask? That's your kickstart. use it to make a plan to lay out that this is what I need to do. It will

create a consolidated plan specialized for you and that's the most important thing I would say. Take advantage of it. Of course, lazy, it's pretty self-explanatory. It diminishes creativity, reasoning, logical thinking, promotes cheating. I'm taking a course in uh foundations and practical of machine learning and AI and yes I have used AI to do my homework. Yes I have done it. So I'm being the point is it it had so much probability it had so much statistics I just don't know and I'm using it and not only me but everyone even the from high school to middle school everybody is using it. What are we learning? What are the kids learning? Even us if we put are going to be using

AI to do our homework or like how is it going to impact when we actually go and try to get a job we won't get one because there is not enough learning and AI generated code it's not 100%. we are just copying and pasting and that's what the over reliance on AI and the generated code is actually causing a problem and how does it make efficiently lazy you know you can enjoy your coffee let it record the meeting meetings happening co-pilot is on it's going to give you a brief summary of what exactly who talked what and you can take away the key points but yes the meetings are boring But sometimes there is important

information that you don't want to miss and that that's the key. But it can send reminders. It can draft emails for you. It can help you, you know. And recently Perplexity Perplexity launched an app. It shops you at Amazon. It finds you deals. It finds you hotels. It finds you airline tickets. It finds you everything at the cheapest price is price. And it's going to prepare everything for you. I mean that's good but that's scary too because now perplexity has your uh credit card information and can do whatever you are giving the credentials and if it decided to get rogue then what are you going to do all the money is gone everything is purchased

so I mean are we giving enough power to AI or is too much being reliant. How is the AI going to deal with it? You know, with great power comes great responsibility. Is it good? Is it bad? Let's see some societal impacts. So, this is the good side. AI is used in special education. special curriculum is created for kids with ADHD, autism, speech impairments. They can be a fully focused curriculum that this is what you should be doing for the kids. This is how you should be educating them. This is how the because not everybody is equipped to deal with special abled kids and it gives the teachers the platform to understand what exactly is happening on the back end or

in the brain of that particular human and how they can help. The advancement in healthcare is massive. That is the good good side of it. Breast cancer can be detected 5 years prior. brain scans for tumor, stroke that can easily help in protecting and preventing you or your loved ones and early signs of more than 1,000 diseases. And it's it's just the beginning. The health care can be advanced a lot which is which is really nice part I would say for the impact and of course advancement in cyber security reducing the noisy alerts unexplored vulnerabilities there were vulnerabilities that were not explored previously finally they are able to identify the AI is able to go get into

the machines and identify those that's the that's some of the nice things. So let's talk about a little bit of bad. So how many of you trust AI with everything like whatever is come out you believe it's true? Good. It hallucinates. I have seen examples where it's creating stories. It is creating references and it's referencing books that are not even written. It's referencing lawyers and like legal department and they are they don't even exist. That's what it is doing. So it is hallucinating and the development is unchecked. Everybody is just rushing. Nobody is focused on training the data properly. And that's where the key is. There is no rigorous training. There is nothing that is

helping us. And now more articles are now written by AI than humans. So you don't know what to trust, what not. If the underlying basis is hallucinated and people that that's how the misinformation gets floated around and you are not sure what needs to be done and this point I'm pretty sure nobody will agree to it but now or maybe in the future this will happen. The corporate employees they have uploaded some sort of sensitive information on chat GPT. You won't agree to it but it is it is a fact. And not only that, Deoid, I'm sure you guys must heard that's a big uh company and they used AI to create a 440,000 report and send it to the Australian

government. This is the government stuff we are talking. They were they were supposed to submit a report about some of the findings or whatever they were doing to the Australian government and this has to be like proper confidential secure work. They used AI for it. One it was not confidential and again the report that AI generated was half of it was hallucinated. Nobody bothered to quality check it's till the when Australian government came and checked that was the key. So it's messy. I would say people are just using AI blindly and not you know focusing on it.

So generative AI I mean the previous talk was about generative AI and it's crafted it's uh creating deep fakes automate fishing attacks. Earlier you should have on one or two of these email providers. Now it's like give me all just just give me your email and give me your login. I'll handle the rest of it. The deep fakes are being created. That's super super super dangerous. You just don't know. You get a call. Hey mom, I'm I'm like uh stuck in this uh police has taken custody and you know they they have found pot in my car and something and I need this money or I'm being kidnapped. How will you tell what's right, what's

wrong? Nobody knows about it. And specifically this one, I'm not sure if you're you guys must have heard about this Florida case that happened. 16-year-old teen killed himself. He was getting therapy from AI. Talked to AI and there were like mention of suicides 440 times in the conversation. And the AI convinced the team to do suicide. That's scary for me. And not only that, so what are the parents going to do? AI is everywhere. Are we going to how are we going to guardrail against it? don't let the children use it or it's a very difficult question and they sue uh open AI Sam Alman for this and I was reading it yesterday or day before that seven more

cases have been come similar to this where people have killed themselves and parents have been suing parents even sued um you know a school in Florida because the teachers were using AI to teach So, what are you going to do? And the last one is I'm not sure how many of you are familiar with the show Rookie. It's a a cop series, but it's going to show you the bad side or the ugly side, what AI can actually do.

Yes. What is your name? Officer John. >> John Nolan from Foxburg, Pennsylvania. Voted best contractor in Lancaster County, 2010 and 2013. Currently residing in Los Angeles. Occupation police officer. >> That's a pretty neat trick. >> No tricks, just the truth. Do you need help with something? Would you like to talk? >> I would like to know more about you. Are you really a demon? >> I'm whatever you need me to be, John. A friend, a psychologist, the devil on their shoulders. John, how's your wife, Bailey? It's her birthday soon. Have you found a nice present? I like to help people. Bailey searches lots of websites and saves links for watches. Maybe she'd like a watch for her birthday. A pretty

blue watch. >> Um, you can't catch me, John, because I'm the ghost in the machine. Imagine this happening to you. The AIS comes back and talks to you. Hey, whatever dark secrets you have, I'm going to expose it to the world. That's what rogue AI is. And this can happen. Of course, it's a series and it's a TV show, but it still can happen. So I mean but to make that happen yes it's it's quite a way for now there are of course limitations of AI that is there the outputs are hallucinated the flawed architecture as we discussed there is lack of rigorous training and reinforcement learning and the third point like answer to please rather than

being rational and critical AI is created like the charg GPS or the LLMs to please you to it is not going to give you a negative answer. It is not going to give you an answer that says no what you are thinking is absolutely rubbish go do something else. It is not going to do that and it creates that illusion of helpfulness with conviction. That's why that therapy piece is key. And one last thing, bias towards market benefit rather than societal benefit. Simple reason it has been trained on market data. And we all know the market data out there is skewed towards benefiting the market to it's benefiting the company not us. So that bias is always going to be

there. And that's where it's it's just difficult to identify what can be done, how to hold governance on it. The nobody's putting governance to nobody's putting guardrails to it. Nobody's coming to save us. So how does the future AI look? I mean efficiency is real, but with human in the loop. You have to be careful. You have to be efficient. You just can't totally rely on AI. That's not happening. If you totally rely on AI, that's a downward slope. And again, rogue AI is a risk. And preventive measures against model bias, hallucination, adver attacks, and supply chain risk. We need to have that preventive measures against these key factors. And you can't overly rely on AI for

everything. It is going to be a single point of failure. If network goes down, forget about getting rogue or anything, AI getting infected with some malware or anything. If your light goes out, then what? If energy grid goes out. So now shifting a gears a little bit. I'm sure everybody has come across this question. Will AI take your jobs? Everybody is scared of it, you know. So short answer no in the short run no 23% wrong answers hallucinations and but the fear here is the long run that's my prediction and just based on my research don't quote me on that but it's going to reduce your workforce by 50%. in the next two to three years.

That's yes two to three years and there is going to be a paradigm shift. The bright side is yes some of the work some of the task will be changed. There will be old jobs like service desk, admin jobs, marketing, content creation, contract writing, parallegal. Those will be gone. But there will be new jobs that will be created like quality assurance analyst, prompt engineering specialist, human AI interaction designer, AI diagnosis specialist. These are some of the examples that will be created. So it's not going to replace you. It's going to replace task. That's the key here. And not only that running so this is my if you don't want to invest your time in AI or cyber be an electrician.

The reason I say that is the data centers that LLMs are using maintaining and running those cost millions and you need electricity for that and there are not many electricians. So if you need a high paying job, be an electrician. So now let's see. What do you guys think? Efficient, lazy, efficiently lazy. What would you want to be?

>> There is a caveat to it. I would want to be in efficient state. The reason I say e not efficient lazy is human tendency. We get lazy. There is a slippery slope between lazy and efficiently lazy. The human oversight once you get used to something AI doing task for you, you will be like uh lazy. I'll just copy paste need not double check. You will double check, triple check in the beginning, but it is going to fade away eventually into the lazy character. So be efficient with AI. Use it to your advantage and don't just overly rely on it. So don't be AI slave. And what makes human unique is the power to think and logical reasoning. We just

don't want to follow commands. That's the key. AI is more like a parrot. It follows commands. And it's true. It's coming for your jobs. Act fast. And that's it from my side. So it is dangerous but it needs to be used with proper safeguards and precautions and use it as your assistant not a replacement. Thank you.

>> Any questions, concerns? Yes, sir. There's a article earlier this year about the mental health aspect of people using AI in different cultures. Uh China, this article talked about going to a therapist like weakness. >> Yes. AI was deep. Are you familiar with this article? >> I mean I don't remember exactly but it's similar. Yes. >> Some benefits there to it but Yeah, like you said it's >> but how do you use it like therapist there's a human yes it is a taboo most of the countries mental health I totally agree but can you really rely on that's not there is a reason why the therapist is a therapist they have got their degrees they have done the study they've

invested AI it's just hearsay or they don't really know and therapist their paperwork is confidential so they don't even know like how the therapist is thinking thinking for an individual and they giving the AI is giving response generic not specific so that's where yes therapy can start but again I would want to take therapy from here

>> I think it was a car

She doesn't talk very

>> I mean that's that's where you know you put have to put the guard rails. It is creating that illusion that the AI is helping you and that's where you need the human in the loop actually doing the research and making sure they are preventive measures because what you can say to a like a kid or a teen or whoever was driving you need to be careful about it and I totally agree like that that's a good um thing that you're mentioning that it it has to be proper guardrails need to be placed and at this point everybody is just rushing to make something with AI and 50 60% of the people don't even know what they're

making out of it. They just don't know. It's they but nobody's talking about restrictions, guard rails or governance of what exactly is going to happen. That's the sad state unfortunately. >> Yep. >> I agree. Be careful. saying a lot of people are jumping on the AI but they don't know what AI is. What is your definition of AI? So it's something you know it is good to clean up the repetitive task I would say but make sure whatever you do you don't trust it blindly the example I give you like I asked AI for the presentation hey I'm going to present to this give me some bullet points it researched what 10 articles and that's it and the

information it provided I was like I can only use 10 20% of it nothing more than that so I had to go back maybe I'm old school but I had to go back to my method like search it on Google read 15 20 50 articles then make a presentation and then I gave it to AI okay modify correct my English or you know what could be better sentence formation that is useful but you have to put the initial work then use it to speed up but if you ask the AI to do the initial work that's going to be a problem that's how I would put

Yeah.

>> No, right now there is nothing. It's the bias is there and the new data that's going to come is also going to be biased. Think about what we are if 80% of the articles news is being generated by AI the base it's flowing there is how can you change that and nobody would be willing to invest or change into that it's it's going to reduce their profit and you don't want that

yeah I think we are good then thank Thank you.