Artificial Intelligence Fear or Future

What in the world is AI?

So, you know the first question most people want to know is what in the world is AI. In today’s day and age, you hear it getting thrown around willy-nilly AI this, AI that, but in reality, the vast majority of people do not have an accurate understanding of what AI actually is. So really, it’s “the science and engineering of making intelligent machines” which again sounds like Silicon Valley fluff. So, really what does that actually mean?

What it really means is software that’s designed to make informed decisions leveraging a massive historical data set, and that really ends up breaking down into two different types of AI. Again, I think when people think AI, they tend to think about this one on the right, which is the General AI, the Strong AI. It’s that human-like intelligence. It’s the one that enslaves the human race, HAL from 2001 A Space Odyssey. The good news for you guys today: that is in a theoretical form. They are at least a decade if not further from that actually happening. There’s many many barriers before that’s going to happen, so don’t stress too much in today’s day and age.

Really what we see today is this one on the left which is the Narrow AI or Weak AI, and what that means is that it’s AI designed for a very specific purpose. This purpose is things like Siri, Alexa, Tesla cars. It’s much more narrow. It’s not evolving, learning, becoming more sophisticated; it’s designed to do a specific task and it’s going to do it.

Brief History of AI

Okay so I want to jump into just a brief history of AI. I promise it will be brief. I know that this can be the most boring section, but I’ll hammer through this. Really the main point of this is that AI has been around a lot longer than everybody thinks. The news cycle really hasn’t picked up on it until recently, but the first beginnings of AI were actually back in the 1950s. Alan Turing authored a paper called “Can Machines Think?”, and the rest of the world just started to kind of follow suit after that. So, in 1955 there was a conference where the term artificial intelligence was actually coined. In 1986 Carnegie Mellon releases their first semi-autonomous vehicle. In 1995, those of you who still own a Nokia phone, know exactly what we’re talking about here. T9 predictive text comes to be. In 1997, Dragon Systems released the first commercial software that did humans speak to text. In 2009, Netflix releases their first search recommendation engine. In 2010, Apple launches Siri. In 2016, open AI publishes their first study on generative models. In 2019 open AI publishes their second research findings on the Chat GPT 2 model, and then Chat GPT launches commercially at the end of 2022. See? Painless. I told you we’d move through that part pretty quick.

AI Is More Common Than You Think

Really there’s kind of two points to that, right? First of all, AI’s been around a lot longer than what the news cycle lets us believe today. It’s also way more common than you think when we talk about the Narrow AI side of things. Things like your GPS navigation systems leverage AI: predictive text, airplane autopilots, social media For You pages (the recommendation engines that are feeding you additional content), and all those online chat bots that annoyingly pop up and ding at you when you go to a website.

AI Key Concepts

So, a few more key concepts that I want to just dive into before we get over the history portion is AI hallucinations and the Black Box problems. If you follow mainstream news, you’ve probably heard one of those two topics, or maybe both of them, thrown out at some point in time.

So, AI hallucinations right. Really, AI they rely on these complicated algorithms that analyze the ways that humans would put words together on the internet. It does not decide what’s true and what’s not. So, when it makes something up, it’s considered a hallucination, which I think is impressive San Francisco marketing for making crap up on the fly.

I apologize in advance for the design of this slide. My design team hated it, but it illustrates a point.

The New York Times ran this study where it was feeding both Microsoft Bing, and Google’s Bard different prompts asking it to reference the New York Times as the source. So, you can see at the top there, that’s the stuff it kicked out and then it cited all this stuff at the bottom. Interestingly enough, the New York Times archive showed that those articles did not exist in any shape or form and have never been around on the internet. So essentially when the AI got painted into a corner, it made something up and decided something that doesn’t actually exist, which obviously makes consumers nervous.

Looking ahead to the Black Box problem, you have your user. They feed a prompt into chat. That prompt then does some stuff, which they call the black box. This is the algorithm that’s making those decisions, reasoning theoretically, we hope, or some kind of Black Box Tech magic that exists. It then returns the answer back to chat, and then it comes back to the user. Then that user kind of gets to ask “hey did you like my answer yes or no.” Then the user’s like “yes I liked it” or “no I didn’t.” Well as many of you are probably already realizing that can cause some issues. The first one is:

How did the Black Box come up with the answer?

Where is the work?

I don’t know about you. I never got credit in math class for not showing my work, but the AIs tend to get a lot of credit without showing their work.

Does the Black Box have a bias?

The short answer is yes. All algorithms are developed by people. People inherently have biases. So people are always wondering, when it kicks back these answers, what are the biases that the black box is producing?

Personally, this one is the one that actually makes me more nervous, which is when the AI goes, “did you like my response person?” Well, most people are not subject matter experts. They have their own inherent bias, and are they qualified to say if it’s a good response? A first-year computer science student versus someone who’s been doing it for 25 years are going to have two different answers and who’s actually checking if the AI is doing the right work is not normally subject matter experts.

Then there’s always this piece. Does the person have a bias or a personal agenda? Are they asking questions of the AI that they don’t like the answers too, saying it’s a bad answer as opposed to it being an objective fact, is definitely in the mix.

AI in the Business World

I want to switch gears now and talk AI in the business world. So, when we talk about AI in the business world, we typically talk about it in the business sense. We’ll jump there in a little bit, but first I kind of want to talk about the consumer side and their reactions to AI in the space. So, when we look at what our consumers worried about, personalized advertising is definitely high up on that list. I know every person here has been having a conversation with their significant other, and all of a sudden, they get ad targeted on Instagram with buying a kayak. You’re like “that’s super weird I’ve never talked about kayaks other than dinner tonight with my wife, and now I’m getting targeted for it”. So, people are definitely worried about that. You know, movie, TV show recommendations, not a super big deal. Neither are music recommendations. The average person likes that.

The job applications are a reasonable concern. There’s a lot of AI going in there, and they’re worried about the bias in the Black Box problem of why are candidates being DQ’ed or why is this job being recommended to me, etc. and so forth. Then chat bots answering questions, people are worried that the chat bots are just giving them what they want to hear and not the actual answers.

Product reviews – this is a big one if you’ve never given any thought to it. AI can write product reviews. So, there are organizations out there that are using AI to write fake product reviews so that the product looks better, and they’re actually improving their sale through rate due to some kind of disingenuous practices on the business side.

Then product descriptions, right? You know, how a computer would describe your new set of sheets. Despite all of those concerns, Forbes did a study and 65% of consumers still feel like they trust businesses that use AI on the very strong side, even higher if you keep the people that don’t really have a strong opinion one way or another.

A better question is, should the consumers actually be concerned? When you think through all of the data that businesses are now able to accumulate around, there is some concern as a consumer. They know your social media habits. They know what you’re searching on the internet. They can scrape things off the internet about you. Cell phones, like Google and Apple, they know a lot about you –where you’re located, where you’re sitting, where your friends are. All of those things they have access to at all the time. I don’t know about you, but my car has built-in GPS. So, Ford knows exactly where my car is sitting at any given moment in time. Then there’s lots and lots more of these DNA sites popping up which we’ll talk about in just a little bit.

A better question is, should the consumers actually be concerned? When you think through all of the data that businesses are now able to accumulate around, there is some concern as a consumer. They know your social media habits. They know what you’re searching on the internet. They can scrape things off the internet about you. Cell phones, like Google and Apple, they know a lot about you –where you’re located, where you’re sitting, where your friends are. All of those things they have access to at all the time. I don’t know about you, but my car has built-in GPS. So, Ford knows exactly where my car is sitting at any given moment in time. Then there’s lots and lots more of these DNA sites popping up which we’ll talk about in just a little bit.

Again, one of my favorite quotes here, “If you aren’t paying for the product, you are the product.” So, Instagram, Facebook, all these social media sites that you sign up for free on. Fun fact, they are making their money somehow, and it’s on you. That’s from Tristan Harris. Look him up sometime. He’s super interesting. He’s done a lot of stuff with Google trying to help keep the ethical side of data gathering in place.

Building a Buyer Profile

So, this is one of my favorite exercises to do with people who are just kind of dabbling into this subject. So, let’s talk about building a buyer profile for me. We already know what my name is, right? So, just from my name, you can find me on LinkedIn, and find out my place of work. From there, you can also find out my job title and the city where I live. So just those three data points and you have a pretty good idea of how much money I make. Check out Instagram and see pictures of me with my lovely wife to find out my marital status. You can see me playing golf and some of my other hobbies and interests. If you have access to Google, you can find out the ads that I click on, the websites I visited, and the reviews that I’ve given, which really gives you a pretty compelling idea of who I am, how much money I have, where I live, and the things that I like to do and I’m likely to spend money on, which is how Google, Instagram, Facebook, and LinkedIn all make heaping sums of money off of your information.

Data is a Business

This goes beyond just social media stuff, because understand data is business. They don’t build all these platforms, give you access for free, for no reason. My favorite example of this is Ancestry. They do over a billion dollars in revenue every single year. They have 40 billion online records: 23 million people in the world’s largest consumer DNA network. Consumer is bolded because if it doesn’t make you nervous, it should. 60 markets globally, records from over 80 countries of origin, and 13 billion ancestral profiles. So, continuing on with additional information here, on average over 1 billion searches are handled by Ancestry servers every single month. Eight hundred million user generated photos, scanned documents, and written stories are stored in their database. More than 3 million paying subscribers and over a hundred and thirty-one million family trees are on ancestry.com today. Ancestry currently manages about 10,000 terabytes of data, including births, deaths, military service, immigration, and much more. This rich collection of information keeps our family coming back for more. If you don’t believe me that’s the link to the corporate website. Every slide that I showed you so far was written verbatim from information taken from that page, so in differing lights, those are two different things. If you showed, up you’re like “wow look at all this great data they’ve got to know about my family”, but in our conversation today you go “oh my gosh look at how much they know about me and my family.” And again, they’re selling that data, building out these buyer profiles to make it easier to target you with more money.

What are business owners worried about?

So, let’s shift gears. What are business owners worried about? Business owners are worried about what you’d pretty much expect. They’re worried about it negatively impacting their customer relationships. They’re worried about the bias errors that we’ve already talked about. They’re worried about it providing their customers with misinformation and having to combat that. They’re worried about privacy concerns. They’re worried about human workforce reduction and what that means to their employees. They’re also worried about maintaining a team of people skilled enough to interact with AI. And then of course this technological dependence that they’re worried that they are not going to have the best people in the future, because they’re going to be reliant so heavily on the AI. Additionally, one in four businesses are worried about losing traffic on their website due to AI. One in three businesses are planning to use AI to create website content. Forty four percent of businesses are planning on using AI to create new content in different languages to reach markets that are historically outside of the English-speaking market. Additionally, 97% of businesses believe AI will help their business.

Fear or Future

So now, heading into our final section here, Fear or Future. So, taking center stage, this graph is going to show the time to 100 million users by platform. So, Twitter did it in five years five months. Facebook did it in four years and six months. Snapchat did it in three years eight months. Myspace did it in three years. Instagram did it in two years six months. TikTok did it in nine months. ChatGPT did it in two months, and threads did it in five days. So, whatever your opinion on AI is, 100 million people signed up for that in the first 60 days of it being publicly available which tells you it is at the middle of the business community.

The Terrifying

So, let’s talk about the terrifying. Government overwatch is a major, major concern for the average person. Thinking about what Google knows about you, the US government, and governments around the world know more. Everyone’s a little bit worried about these massive tech monopolies: Google, Facebook, Instagram, etc. having access to all your personal data. Fake and deceptive news; that is a big issue where if you’re writing your news articles leveraging the AI, it’s making up sources. It’s got its inherent biases, so again be mindful of that. Social media addiction, they’re talking about how the average person spends on average on TikTok 90 minutes per day, which is a little bit cringe when you think about it. Copyright and plagiarism issues, those who have created digital art or writing, they’re training these AIs off of massive data sets. Massive data sets being publicly available art and text information. Then obviously job displacement, right? You’re worried about being out of one sooner than later. So, one of the questions that our clients often ask us is how do we regulate this.

How do we regulate this?

Do we trust in our government protectors? Do we trust in the Fortune 500? Do we just create a bunch of laws and regulations and rules around that? It causes some bigger challenges, and it’s worthy of a discussion.

So, the main issue with legislation, besides Congress, is data privacy laws. They typically cause two main issues: they increase the barrier of entry for startups in smaller companies and they centralize your data with monopolies. So again, people often don’t understand what the ramifications of that are. The European Union was rolling out GDPR for consumer data privacy. 8.9 billion dollars were spent on that just from Fortune 500 companies and the UK’s largest 350 companies. I mean you think about the amount of money that they’re spending on this, your average startup can’t compete, which keeps them out of the market, which keeps pressure off the monopolies, which essentially continues to allow them to do whatever they want. So, when you don’t have any small businesses coming in, you have data that gets centralized in monopolies. What’s the problem with that? Well, they become a larger target for people to hack. All the data is now in one place.

Imagine if JPMorgan only had one place that kept the money. It would be a vault that people would be interested in getting in to. Monopolies also have an impressive ability to settle when they make mistakes. They have a lot of buying power, a lot of lobbyists, and they typically settle for pennies on the dollar, of, realistically, what they should be paid.

Then innovation is stifled by creating those barriers to entry. So, a good example of this is the Equifax data breach. Equifax being one of the three major credit bureaus in the United States. Their data breach exposed the personal information, such as Social Security numbers of 147 million people. My guess is, if you’re watching this, you’re one of them. On top of that, they settled for 425 million which sounds like a lot of money until you really do the math and realize that it is $2.89 for each of the infractions that they actually had.

Personally, I would let them keep the two dollars if I could keep my social security number private for the rest of my life. Now, it’s not all doom and gloom and rain showers. There is a lot of extraordinarily promising things that are going on because of AI, in medical research alone. Again, they train AI on large datasets. They’re leveraging AI to find predictable patterns in early onset Alzheimer’s. They’re doing work on the cancer side of things. With this information, they’ll be able to find data trends and early analysis on serious diseases significantly faster than they ever had at any point in human history, which is an amazing first step.

The extraordinarily promising.

My world, software development. Honestly, the AI is getting fairly good at writing outside code, which is helpful to make software developers more efficient, and they get to spend time on the things that actually matter. Education, there’s a lot of concerns around copyright/plagiarism, but there’s also a lot of promising things. The AIs are going to be able to provide real-time feedback when you’re doing a math problem. You can say “hey what am I doing wrong,” and it will be able to help walk you through and understand the information. So, if you truly use AI to build an education and not to get the assignment done, it’s a massively powerful tool.

Global infrastructure, right? There are major engineering firms that do work all over the planet that will now be able to leverage AI to help build better infrastructure in all these up-and-coming countries at a much faster rate than they ever have before. Travel and transportation, we’re already seeing. The last flight you boarded, most of that was piloted by the computer. You’re going to start seeing that with cars/semi-trucks, so that the issues in the supply chain will become less and less frequent, because they’re going to have computers running the show here before too long. Then obviously, in the agricultural space being able to determine based off weather patterns that you should or should not be watering your lawn, should or should not be watering crops. Really, the ability to create these massive datasets will have a massive impact on the agricultural space as well.

AI in the year 2030

So, looking ahead to 2030, they’re estimating that AI is going to contribute 15.7 trillion to the global economy. They think over 2 billion jobs could be displaced in the global economy over the next several years. Up to 47% of jobs in the U.S could be impacted by computerization, and 70% of the global GDP is expected to be in China and North America. Again, those are scary sounding things. They’re not as bad as you think they are. One thing is for certain. AI is not going away. So, at the end of the day what does it all mean?

What does it all mean?

AI is going to change the way we do business. There is no doubt about it. It already has. It’s going to continue to do so, as it continues to get more sophisticated. AI is going to open the door for major regulations around data privacy and what companies can and can’t do with it. Then AI, I mean it’s still got a long way to go. It is not perfect, as we talked about with the Black Box Theory and those AI hallucinations just making stuff up, but it isn’t going to go away anytime soon.

I run a business. What should I do?

So, you might be asking yourself, “Well I run a business. What should I do?” and in my opinion invest your time in the high impact areas. These are things that require a lot of time. They’re usually text based, you know, things like call centers, writing emails, legal services, document writing, all the authoring of books, and things like that. Insurance has a lot of text and forms. Healthcare, some software stuff, translation, and documentation are all areas that have high time requirements that AI can help leverage to make this a more efficient process.

Again, focus on the high time activities: blog posts, templating, emails, generating quiz questions, chat bots, basic translations, proof reading, coming up with ideas, basic branding, API Integrations, all things. Some research, depending on what you’re doing, can be done, and then we kind of start talking about complex code, problem solving, and base business logic are all things that fall into that category that I would avoid.

So, this graph, you see the ones in red those are the ones that I don’t think AI is quite up to solving yet. The API integration is in blue there again. That’s my world. It does a great job with some, a pretty terrible job with others. It really just depends, but the areas that are highlighted in black are all things that AI is doing fairly well today and can save your business lots of time and money by automating those processes or at least semi-automating those processes leveraging the AI that exists today.

AI Do Nots

Some do’s and do nots. So right now, don’t leverage AI for personal or business finances. Loading that information in today, again, it’s getting stored on some server some place with very little regulation. We don’t really know what they’re doing with it, hence the Black Box problem. So don’t put your financial information into the AIs, yet. Don’t send any confidential information such as birth dates, addresses, and Social Security numbers. Even though as we’ve already talked about, your social security number is already owned by a bunch of people thanks to Experian. Upload images of you or your children. They are training the AIs to generate images based off the images they can find. If you don’t want you or your children having their images up there, don’t feed it to them. Also tell your relatives do not upload them into ancestry.com, because you don’t know where they’re going either.

Don’t take AI at its word or as news. Again, some experts say, “Don’t ask it a question you don’t already know the answer to”. I think you can go a little bit beyond that, but don’t lose the curiosity with the information that it kicks back out. AI’s not your doctor, so don’t treat it like that. That’s not what you want to be doing. You know, WebMD is already scary enough. Don’t let AI create monsters. If you have a medical condition, go see a doctor.

Do not use AI as legal counsel. While it is helpful in generating templates and employment agreements, again, always have that reviewed by an expert. The intent behind AI is not to replace your lawyer, it’s to make them more efficient, so they can spend time doing the value-added things, so you’re not spending $400/hour having them write up a memo for you. Instead, you’re paying them for five minutes to review the memo that the computer generated.

Ask Yourself

So, ask yourself: how can I differentiate my business with the help of AI? My business, we do software consulting, right? Obviously, people are like, “Well can the AI replace some of your job?” and I ask myself these questions: How can we improve communication with our clients? How can we provide them counsel, guidance, advice that the computer is not going to give them? Justification for doing it? The strategy piece is now a lot more center of our business than it ever used to be. So, if you run a business that you’re worried that AI is going to replace you, figure out ways to improve your communication.

Additionally, how can you just become more efficient, right? For us, we leverage AI to write the boring code that’s simple, repetitive, and lets us focus on the high value code. The ones that are doing the business logic and the really complicated things that you should be paying your software developers to do, not to write the basic Integrations.

Again, get back to spending time on high value tasks. There’s a lot of things out there in the business world that create a lot of noise, but the high value tasks are where you should be spending your time. What can AI take off your plate to let you focus more on your time on, is the question you should be asking yourself.

Conclusion

This is a deep topic for sure. There’s a lot of information out there. We’re here always to talk, to provide advice, guidance. If you have a project that you need help with, if you need some of that, let’s chat. I’m more than happy to chat anytime.

You can find me at the Definity website, and there’s the QR code to our contact form. Thank you all so much for your time today. I appreciate it. Once again, my name is Shawn Barber, the CEO of Definity, and thank you so much.