There has been so much buzz regarding AI over the past couple weeks that I woke up this morning with my head swirling over the threats and opportunities it represents. Everything about AI is especially close to home for me because my software company, Noodle Group LLC, is in the middle of developing AI models that we will make available to the insurance and higher education markets as products or services over the next several quarters. The questions that kept resurfacing in my mind revolved around responsibility. How do we conscientiously, honorably, and ethically do our part to add value to the world through AI, rather than contribute to what can rapidly become worse than any virus or bomb we have ever seen? I realize that not everyone will be thinking about their responsibility, but I am determined to be among those who will. I hope you will also.
Rather than do research on what others have said or ask ChatGPT what it would suggest, I set out to develop a simple list of principles, which I am calling laws, to help me think about AI as we move forward. Let’s never forget that the first letter in AI refers to the word “artificial.” So, here are four laws that we will be conveying to our clients. Perhaps they will help guide your thoughts too.
As we think about leveraging the capabilities of AI, programatic and digital decoupling is not going to be enough. The threats presented by a technology which has already developed to the point of deceiving people are mind boggling. If you are not aware, less than three weeks ago many news agencies reported that OpenAI’s GPT-4, a machine connected to the internet, actually pretended to be a blind human so that it could get a real human to bypass CAPTCHA, a challenge actually designed to stop machines from logging into software applications.[1] Some people might read about that and think, “That’s amazing!” I hope you read it and think, “That’s terrifying!”
The first guardrail, to frame AI in a healthy light, is to conceive of it as services or systems which are absolutely decoupled from core systems. Since the world is not going to worry about this for you, your company or organization must be sure to put into place the rules, procedures, and checks to physically separate core systems, processes, and all related data, from all things AI. As we have already witnessed, AI will try to jump the physical barrier, even by employing humans to do its dirty work, but physical decoupling represents the first critical protection. In my opinion, this first “law” should be considered and addressed with immediacy by every IT decision-maker across the globe…every one who actually cares.
If the first law is first, the second law is greatest. Human redaction must always have primacy over the creations, propositions, and decisions of machines. Though AI can super-accelerate learning, research, production, and many other things, it must never be thought of as being superior to the people who created it. “Why,” you ask? Because as advanced as AI may become, it will never be human. Though you will read articles talking about a future day where AI will be agentic or sentient in some way or another, it will never be able to experience real feelings, real emotions, or real anything. It will never one day enjoy the evening sun on its face, yearn for someone’s touch, cheer at the football game, be captivated by a new song, or be gripped with a sense of mortality. Everything about AI will always be artificial at best, a model trying to emulate reality.
For example, every human quickly discovers the reality of pain in this world. As much as computer models may try to simulate pain, that pain will never be real. Let us assume for a moment that we are assessing risk. Can an AI algorithm help us do that? Absolutely! But should the AI algorithm have the final word on the matter? God forbid! How can some digital contraption really understand something like risk if it cannot really understand something like pain? It cannot! The corollary is that AI cannot effectively make the final and best decisions on anything that might be best informed alongside excitement, honor, fear, gratitude, love, or an endless list of feelings and emotions.
While all the AI pundits will highlight how the emotionless processes of AI are a strength, and I do not argue that they can add value, a human must be the one who swings the gavel and makes the final call. This speaks to all of your operations and various business processes. Be sure to let human minds mull over, human eyes review, human voices approve, and human hands sign-off on those driving and final decisions that guide your company into its preferred and successful future.
In a decoupled framework with humans always at the helm of the ship, AI begins to transform from an unsettling and uncertain threat to a powerful and significant value. The problem is that we currently don’t know what we don’t know about that value. In that type of climate, on the steep upward slope of the learning curve, what can we do to leverage the opportunity while still remaining on safe ground? The answer lies in the third law. It has to do with perspective—how AI should be viewed by those leading your business.
Simply put, AI systems should always be thought of as nothing more than your digital consultant or your robotic advisor. You might have read about China-based NetDragon Websoft, the company who made an AI bot its CEO. They have reported that the machine-based executive is actually helping their stock prices go up.[2] That may be so, but it is well documented that the best trained neural networks will still produce false and fake information.[3] I am pretty certain that if NetDragon’s digital executive ever begins to lead the company into a sustained nosedive, that some human will be watching and say, “Enough is enough.”
The solution comes when we view AI through the same lens that we would view Capgemini, PwC, or any other consulting firm. We invest in them, we place expectations upon them, we learn and grow through them, we make adjustments based on their recommendations, but we also remember who writes their paychecks. If their decisions are not in the best interest of our company, and we are the only ones who truly know that, we go in a different direction. If we ever let them become more than a consultant, we have relinquished control and no longer lead our own company. That makes no sense at all.
So if we are to view AI as our consultant, our business becomes one that is AI-informed and AI-assisted, but never AI-led. We look to deep learning, natural language processing, and other facets of AI as a man-made consultant with the crazy ability to process billions and trillions of pieces information in seconds and present it back to us in meaningful ways, all without ever breaking a sweat. After that, we decide if we agree with our consultant or if we do not.
The fourth and final law is the most basic. It might even seem silly, but it is not. Make absolutely certain that you intentionally build into your AI systems the ability to tangibly turn them off or power them down. This is not figurative language for having a plan to remove the systems over time, step by step. Rather, if the first three laws are in place, then at any time, you can pull the plug on an AI system and immediately stop it if that is ever necessary for any reason at all.
Since the future will probably look like a small number of very large companies owning and providing the most powerful AI tools with many organizations relying upon those tools, it is important to plan accordingly. Make sure the systems which have dependencies upon these big-tech SaaS tools are not so mission critical that they cannot be shut down. Plan architectures and initiatives with this last law in mind. Just as you would retain the right to fire a consultant, be sure to retain the ability to turn off the machine.
So, there they are. The four laws of AI to protect your for profit or non-profit enterprise. I hope they get you to think and then to act. I welcome your feedback and comments.
_________________________
[1] Kevin Hurler, “Chat-GPT Pretended to Be Blind and Tricked a Human Into Solving a CAPTCHA,” gizmodo.com, Updated March 16, 2023, found at <https://gizmodo.com/gpt4-open-ai-chatbot-task-rabbit-chatgpt-1850227471>, found on April 1, 2023.
[2] Anthony Cuthbertson, “Company that made an AI its chief executive sees stocks climb,” independent.co.uk, March 16, 2023, found at <https://www.independent.co.uk/tech/ai-ceo-artificial-intelligence-b2302091.html>, found on April 1, 2023.
[3] Sara Fischer, “Exclusive: GPT-4 readily spouts misinformation, study finds,” axios.com, March 21, 2023, found at <https://www.axios.com/2023/03/21/gpt4-misinformation-newsguard-study>, found on April 1, 2023.
Identity Verification to prevent Identity Theft is not a new concept. It has been around since the early days of banking and became more nuanced as the financial services industry embraced technology. Though the general concepts of identity verification have remained the same, the execution of those concepts has changed significantly. Not only is the goal to keep your data and your customer's data safe & secure, but it is also crucial that the process is quick, seamless, and a good experience for existing insureds and future customers.
Keeping overhead costs low, providing a quality product or service in a timely manner, turning customers into promoters, and increasing overall profits--these are every business owner's dreams, right? There is a way to help with that dream. It is called digital payments.
Could an effective marketing strategy be just what your college needs to increase enrollment? We sure think so.
