Artificial intelligence offers tremendous possibilities.

Because of AI, we’re able to do unbelievable things. We can mimic health conditions and safely explore the interactions of pharmaceutical interventions. Exploring the management of microclimates is possible due to a series of complex algorithms. AI makes it possible to predict production levels, find instances of fraud, and take on routine tasks.

AI can’t do everything we’re hoping it will do, at least not yet. If we want artificial intelligence to become a conscious replacement for humans, there’s more it will have to learn.

For now, here’s what it can’t do.

Engage in ethical decision-making

Artificial intelligence doesn’t engage in ethical problem-solving. It’s not even worried about being politically correct. Only people do can that. AI can run algorithms and ferret out possibilities, but in the end, it’s the people analyzing the results who make the final decisions.

Additionally, AI will only be as ethical as the programmers writing the algorithms.  Programmers influence every bit of technology we use: our smartphones, computers, and the software we use. The artificial intelligence they are building makes decisions based only on what it is taught. 

The Association for Computing Machinery (ACM) recognizes the impact of ethical programming in computing. Deceitful coders could write algorithms that subvert AI goals or cause machine intelligence to choose incorrectly. We have only to look at Uber’s self-driving cars and Amazon’s Alexa-initiated purchases for example.

AI cannot differentiate between what it could do and should do.

Provide emotional support

Worry about artificial intelligence taking away jobs is misplaced. While AI may take on some of our more routine tasks, it will not take on roles where people must nurture and support each other. The mindfulness required in teaching goes beyond the capabilities of AI. Educators bring a socio-emotional connection to the classroom as they connect with their students.

So far, we haven’t been able to teach the critical components of emotional intelligence to our machines. These include active listening, demonstrations of empathy, and the use of self-control.

AI mimics human emotion. It has not internalized it.

Generate humor – and laugh

Knowing what’s funny requires more than an understanding of language. There’s also a set of social cues one must be able to interpret.  Humor is based on inconsistencies. The incongruous juxtaposition of verbal communication and visual gestures makes people laugh.

Artificial intelligence can recognize a joke. AI assistants can tell jokes when prompted.  However, understanding the humor is beyond AI comprehension. They don’t intuit incongruities.

The take-a-way

AI has made tremendous leaps in its ability to appear human. It has taken on mundane tasks we don’t want to do, like vacuuming and sorting. It can predict financial models and diagnose medical conditions. These are, however, only manifestations of our own design.

Until machine learning can make ethical decisions, show emotion, and laugh at itself, it won’t measure up to human abilities or consciousness. In spite of the hope it brings, artificial intelligence still has a long way to go before it can do it all.