Microsoft's AI boss is right: sentient AI fantasies aren't just impossible, they're irrelevent

Microsoft AI CEO Mustafa Suleyman's opinions on AI's shape and development carry some weight, which is why it felt like a breath of fresh air to hear him say that AI cannot achieve consciousness and that pursuing it misunderstands the point of the technology.

The idea of Frankenstein-ing sentience into AI chatbots gets a lot of buzz, but Suleymans' comments at the recent AfroTech Conference dismissed the very idea of artificial consciousness as starting from a false premise.

“If you ask the wrong question,” he said, “you end up with the wrong answer.” And, in his view, asking whether AIs can be conscious is a textbook example of the wrong question.

Pushing back on the breathless speculation about artificial general intelligence (AGI) or claims that ChatGPT has achieved self-awareness is something more people in AI with some authority on the subject should do.

Not that Suleyman is against building new and better AI models. He just believes it's better to focus on making AI into useful tools for people, not pretending we're nurturing a digital Pinocchio into a real boy.

The distinction between AI performing well and AI being aware is crucial. Because pretending there's a spark of real self-awareness behind the algorithms is distracting and possibly even dangerous if people start treating these fancy auto-completes like they're capable of introspection.

'Smart' doesn't mean 'thinking'

As Suleyman pointed out, it's possible to actually see what the model is doing when it mimics emotions and feelings. They don’t have hidden internal lives. We can watch the math happen. We can trace the input tokens, the attention weights, and the statistical probabilities as the sausage gets made. And nowhere in that pipeline is there a mechanism for subjective experience.

Dwelling on the mistaken belief that simulated emotions are the real thing is a waste of effort as it is. But when we start responding to machines as if they were human and anthropomorphizing them, we can lose track of reality.

People calling a chatbot their best friend, therapist, or even their romantic partner isn't more of a crisis than treating a fictional character or celebrity who's never met you as an important part of your life. But having a true breakdown over a tragic end to your favorite character in a novel or changing your life to match a fad promoted by a celebrity would be rightly considered concerning. The same worries should arise when a user starts attributing suffering to a chatbot.

That’s not to say it isn’t useful. Quite the opposite. And a little personality can make tools more engaging, more effective, and more fun. But the focus should be on the user's experience, not the illusion of the tool’s inner life.

The real frontier of AI isn’t “how close can we get to making it seem alive?” It’s “how do we make it actually useful?”

There’s still plenty of mystery in AI development. These systems are complex, and we don’t fully understand every emergent behavior. But that doesn’t mean there’s a mind hiding in the wires. The longer we continue to treat consciousness as the holy grail, the more the public is misled.

It would be like seeing a magician pull a coin from your ear and deciding he's truly conjured the cash from nothingness and is therefore an actual sorcerer. The trick is now an over-the-top misunderstanding of what happened. AI chatbots pulling off sleight of hand (or code) is a good trick, but it's not really magic.

The best business laptops for all budgets

➡️ Read our full guide to the best business laptops
1. Best overall:
Dell Precision 5690
2. Best on a budget:
Acer Aspire 5
3. Best MacBook:
Apple MacBook Pro 14-inch (M4)