Judgment Day is coming… for AI

Artificial intelligence has great potential in predicting outcomes. While AI can predict the likelihood of precipitation, it most likely won’t help you dress or prepare for inclement weather.

That preparation involves more than the probability of rain or snow. Humans make extremely subtle judgment calls on what to wear based on their preferences, intuition, or life experiences. Current AI lacks these attributes and nuanced thinking.

The key distinction here is that AI can make relatively accurate predictions based on all known information very quickly, but it still lacks the judgment humans have.  In fact, having ALL the information can be a handicap.

Because of generative AI and large language models (LLMs), AI can do amazing human-like things such as pass a medical exam or an LSAT test.  But would you trust it to serve as your physician or attorney?

AI is a tool, not an expert. 

The irony is that generative AI (genAI) knows just about everything about nearly every subject because it can instantly read and curate everything that is readable. But in the end, it is in fact a statistical program designed to produce the most likely outcome, which is most often correct. 

Most often, but not always.

When asked to identify the greatest leaders in the history of the world, Microsoft CoPilot responded by citing a number of lists from the likes of the BBC and Fortune that ranked Mahara Ranjit Singh as the greatest leader due to his progressive political, social, and cultural policies. Others cited similarly were Mahatma Gandhi, Nelson Mandela, and Martin Luther King Jr.

Some leaders were conspicuous by their absence — no Roman emperors, United States presidents, or renowned military leaders were on the list.  Neither were those who are typically credited with establishing or preserving democracy.

Clearly, what CoPilot did was capture and curate several media lists, which it ranked high on authority. But it applied one set of criteria – peaceful resistance for social change – and synthesized this short list. The point is: the AI was simply citing (and taking a very narrow view of) human judgment because it lacked its own.  But judgment day is coming for AI.

Calling AI “artificial intelligence” implies it has human-like intellect.  Perhaps it should be considered “artificial knowledge,” for the data and information it collects and the wisdom it lacks.

From healthcare diagnostics to financial forecasting, AI’s potential to enhance efficiency and accuracy is undeniable. While AI can predict outcomes based on data, it lacks the graduated and often punctuated understanding that human judgment involves.  Humans bring to the table their senses, experiences, and ethical considerations – factors that AI, as of now, cannot fully replicate, especially in a combined and learned fashion.  This gap underscores the importance of maintaining human oversight over AI systems, ensuring that decisions are not only data-driven but also ethically sound and socially responsible.

AI knows too much about all data but very little about life.  It knows everything that is publicly and electronically accessible.  And since it knows too much, it reverts to the mean.  It finds the most common answer, essentially averaging out all the opinions and facts it has ever collected.  Spawning a million robotic genAIs imitating the average intellect on the Internet is not going to further society or solve any complex problems.  It will actually drive us to a form of perfected mediocrity.

Reaching the level of human discernment comes from the life experiences people have that AI has not been afforded.  Truly intelligent AI will emerge when we form one that involves learning over time and begins to understand what it means to be human and can act in service as a tool.  The Internet is a tool.  Expecting AI or the Internet, for that matter, to have all the answers is naïve. Humans are a product of our education, our experience, and our environment.  These all must be part of the development of a true AI.

What I hope for is a future in which artificially intelligent and prescient beings can be created that will be able to experience life or a close simulation of it.  Beings that take courses, work on project teams, and act as part of an ecosystem with other prescient beings, making trade-offs and coexisting with others.  Beings that face obstacles, scarcity, limitations, and consequences must apply insight and intuition to overcome or learn through repeated failures, often in unfair situations with imperfect conditions, just as we do.  These will be more trustable, intelligent, nuanced and ethical creatures.

The only way will be to raise them through a digital version of our life-like experiences living through middle school, high school, good and bad relationships, growing a family, being dependent on chance and others, or even building a business.  These are the things that give people the judgment to make the right, not the average, decision as a parent, boss, teacher, friend, and human.

I don’t think we will be able to achieve this for some time. Until then, let’s make sure we do not put AI in charge of making critical decisions that require judgment and finality.  Let’s recognize there is a big difference between stupendous artificial knowledge and the true intelligence and wisdom that drives human judgment.

Let’s look forward to a world where we can train bots through some realistic experiences and environments, so they independently recognize significant leadership, achievements, failures, and all the good and bad traits of humans to simulate morality to make the “right” decisions.

I am not advocating for artificial humans that can replace us.  As an AI leader in the customer experience space, I want to make sure we responsibly leverage AI as a tool with a critical eye that enhances human capabilities. 

AI is a powerful tool that will accelerate and improve human performance and experience by distilling vast amounts of information and predicting outcomes that humans can then judge for action.  AI is a great tool, if used safely and properly.