Not long ago I was at a Government Conference on data analytics, finance, and economic development. At the Conference, like all conferences, there was a very large area for venders, and all of them were hawking the latest and greatest in AI Advancements. Needless to say, as a Senior Executive, I was not impressed; each pitch centered on buzzwords, marketing slogans, but no real answers concerning what they are claiming as their advances in AI and Government.
Let me start with this, I am neither pro nor con AI; I have seen it used as a marketing tool to impress consumers thinking they are getting the latest and greatest in technology with increased pricing. You can even get a coffee maker with and AI Assistant, e.g. JavaGenie AI, with an AI Assistant, ranges from $200-$800. Think of it this way, does a coffee maker really need to come with an AI Assistant? After all, you put the cup in the designated area, you add the beans, grounds, or pod in the machine, you select the brew level and strength, you add the water, you set the timer, and it brews the coffee. Once done, you add the condiments of your choosing and enjoy. It does comes with another App for your phone, is that the AI Assistant?
Here is something very important to remeber, true AI does not exist. True AI, or Artificial General Intelligence (AGI), does not currently exist. What many are selling as an AI advancement in our industry, are simply advanced algorithms. Statements like, Empowered leaders rely on our AI-powered analytic tools to gain insights, blah, blah, blah, are marketing statements designed to make you feel if you are not using their tools, and paying big money, you will never get the results you desire.
Current AI technologies, machine learning and narrow AI, are widely used and have made significant advancements, they are limited to specific tasks and do not possess general intelligence. I am not saying, running data analytics through an advanced algorithm is not valuable, it is. It can help in team engagement, help identify performance trends, assist in leadership effectiveness, share real-time dashboards on needed information, help leadership and others see the effective nature of their work. It can help you develop a clear picture of next steps. Just keep in mind, experts agree that actual artificial intelligence, in the sense of AGI, remains a theoretical concept and has not been realized, and human-made technology does have potential, it also has limitations.
The use of AI in decision making processes can lead to several pitfalls. These pitfalls can undermine the effectiveness and reliability of any decision you make. Common issues you need to be aware of are:
The Lack of Transparency and Explainability: Not long ago I had a conversation with one of our legacy venters. I wanted to know details behind how the system was calculating our data; the data seemed off. I was assured the calculation of our data was correct but could not share details due to the proprietary nature of the algorithm. I told him, the assurance of correct information does not explain how the data is calculated. I soon received the needed information and quickly moved to another vender. It turned out, they base their algorithm on Florida Laws [we are not in Florida]. I found our data [PI] was sold to third party organizations and resold to other organizations, a violation of our State Laws. It seems no one asked the right questions for over 7 years.
AI models are often complex, difficult to interpret, shrouded in proprietary language, challenging to understand, and piled high in BS. I always recommend having two to three people who can ask questions you may miss, and vender must be clear in how they reply. A lack of transparency and explainability WILL lead to mistrust and make it harder to identify biases in the decision making process.
The Horrors of Biased Data and Algorithms: Because the AI model used was based on Florida Laws, it was biased; it did not reflect the information we needed in our State. Biase, models will learn, focus and replicate those biases. This will only lead to unfair or discriminatory outcomes. Flawed algorithms will lead to poor decisions and perpetuate existing social inequalities and may even create new ones.
Poor Data Quality Issues: Garbage in, Garbage Out was once the mantra of everyone involved in analytics. Today, that mantra could be Good Data In, Garbage Processing, Garbage Out. While data quality, missing values, incorrect formatting, or outdated information is still an issue, how the algorithm processes the data will significantly impact the accuracy and reliability of the information received. Lead to incorrect decisions, recommendations in high stake outcomes.
There are Always Security Risks: All systems are vulnerable to cyber-attacks. Data breaches, and other security threats will compromise the integrity of the data collected and the decision making process. Security is not only an issue within the organization, but [as I found out] they can be found in the outside vender. A breach in security can have serious consequences, to you, your organization, and the community you serve.
We Can Develop an Overreliance on Technology: I am by no means anti-tech. But I do believe, when we over really on tech, we lose human judgment and critical thinking; we ignore the gut feeling in favor of following the tech blindly. Overlook the importance of context and possible risks associated with our decisions.
When trying to answer the question of whether AI is just a glorified algorithm can be complex and rather subjective. Yes, algorithms are a component of AI and are just one part of the equation. When using AI, we need to not negate the human factor, the gut feeling you get when you look over the data, and teams to ask deeper questions. Ask questions, think as if the box no longer exists, and see the interconnected nature of what you are trying to achieve.
About the Author
