Hassan Taher: 8 AI Pitfalls We Need To Overcome

What are the AI pitfalls we need to overcome? The artificial intelligence debate is one of extremes. Some imagine a Terminator-esque extinction event brought about by intelligent machines. Others see only promise.

AI expert and esteemed author Hassan Taher believes the truth is a bit more nuanced. He recognizes the great potential of AI. Every day, he sees how it’s helping people across industries work faster and smarter.

But he sees AI as the double-edged sword that it is. People are already feeling the impacts of its pitfalls and the potential of AI in their daily lives. “Like every major technological advancement in human history, AI presents a duality — the brilliance of its potential matched by the shadows of its pitfalls,” Taher wrote in a blog post.

Hassan Taher shares his expertise and takes a deep dive into the eight AI pitfalls / potential issues humans have to be on the lookout for as AI inserts itself into daily life.

8 AI Pitfalls We Need To Overcome

1. Trying To Automate Complex Moral Decisions 

Should self-driving cars prioritize the driver over a pedestrian who casually walks into the street? What about a group of pedestrians over one driver? Would that change the decision-making process if that pedestrian is pushing a baby stroller? Humans make split-second moral decisions when faced with an inevitable car accident. AI gives society time to deliberate over the fairest and best solution.

It’s not proving easy. However, one thing is for sure: AI is advancing faster than legislators can keep up. Protecting the public without stifling AI’s potential will be a team effort between the public and private sectors that will need to answer these questions proactively rather than leave them for a machine that has no sense of morals, ethics, or human experience to decide.

Wrote Taher: “With great power comes great responsibility. As we embrace AI, we must also grapple with the challenges that arise.”

2. The Question of Accountability

If an AI-guided robotics surgery goes south, Hassan Taher asked, “Who’s accountable if things go awry — the physician, the programmer, or the machine?” 

Back to that self-driving car for a moment: If that self-driving car prioritized a pedestrian over the driver, could that driver’s family sue the developer of the self-driving technology? Couldn’t the pedestrian’s family do the same in the ultimate Catch-22? 

In both cases, society must demand transparency and accountability for the ethical application of AI. It must establish who is accountable and to what extent a human is held responsible when something goes wrong.

3. Historical Biases Can Perpetuate Inequality

Historically, data has been collected in both known and unknown biased ways. If AI makes decisions based on partial data, the results will be no different. It’s critical to identify potential bias and work to negate it. In doing so, society can potentially eliminate bias in decision-making with AI. Hassan Taher warns, “Beyond data, algorithms, often seen as neutral, can also carry underlying prejudices, which can be more insidious as they’re harder to detect.”

4. Lack of Transparency Breeds Distrust

In an opaque AI industry, those using (or subjected to the use of) AI don’t know how AI is reaching certain conclusions. That’s not only dangerous — it increases distrust, hindering the effective development and use of AI. It will slow developers’ ability to fix the problems.

For example, if an AI system gives an extended prison sentence to one person while being lenient to someone else, AI may be basing it on the likelihood that a certain person will commit a crime again. But is that recommendation biased by race, income, marital status, eye color, type of car driven, or another factor that would be unethical to use in this decision? AI must be transparent enough that someone can evaluate how it’s making decisions and ensure its judgment isn’t based on things that have no real impact on someone becoming a repeat offender.

5. Danger in Assuming AI Knows Best

With so much data, it’s easy to assume that AI is making the best and fairest decisions. However, this isn’t something that should be assumed. Say that doctors use AI to identify the best candidate for a kidney transplant to prioritize the waiting list. 

One could assume AI will always make the fairest decision. And the truth is that AI can make the most suitable and unbiased decisions. But it all depends on what information AI is given.

6. Overestimation of Potential Can Lead to Disillusionment

This will be hard for some people to hear. But AI isn’t intelligent … yet. It predicts outcomes based on historical data, albeit more data than any human could process to make a decision.

According to Taher, “Striking a balance in our expectations of AI is crucial. Overestimating its present capabilities can lead to disillusionment while underestimating its future potential can catch us unprepared.”

7. The “It Took My Job” Mentality

“The AI revolution, like past technological shifts, will redefine the workforce,” Taher wrote. “While certain roles may become obsolete, new opportunities will emerge. Adapting to this changing landscape is a societal challenge.”

From eliminating the steamboat coal shoveler to the horse and buggy driver, humanity is constantly learning to do more with less human labor. But as Hassan Taher discussed, if there’s one thing that history tells us, it’s that each time people start to think, “It’s going to take our jobs,” technology creates more jobs than it replaces. 

Moreover, the new roles are often safer, more fulfilling, higher-paying, and less health-destroying. With that said, they may require the development of new skills, so they do take some adaptation on the part of humans. 

8. Inequitable Benefits Distribution

Countries with infrastructures to use AI are likely to use it to improve the lives of their citizens, further increasing wealth and their status in the global community. This can widen the divide between prosperous and less wealthy economies. 

To mitigate these risks, developing economies must proactively develop infrastructures that enhance their ability to use AI. But here’s the thing: AI has the power to accelerate innovation. So, as they shift in the right direction, the AI will support their efforts. This can change the power and wealth dynamics between nations for the better. When wielded for good, AI could reduce, rather than widen, wealth and access gaps. 

This falls in line with a sentiment Hassan Taher has shared time and again: “I believe that AI technology has the potential to bring about significant positive change in the world, but many people are hesitant to embrace it fully. While some may disagree, I believe that with responsible use, AI can actually make the world a better place for everyone.” Hope the importance of these 8 AI pitfalls are clear now.