Can a machine ever truly feel pain, joy, or fear?

Microsoft’s AI chief, Mustafa Suleyman, says the answer is a clear no and he believes it’s time to stop even asking that question.

“I don’t think that is work that people should be doing,” Suleyman told CNBC at the AfroTech Conference in Houston this week. “If you ask the wrong question, you end up with the wrong answer. I think it’s totally the wrong question.”

Suleyman, who leads Microsoft’s AI division, has been one of the loudest voices pushing back against the idea of “conscious AI” the belief that machines could one day think or feel like humans.

The Line Between Intelligence and Emotion

In his 2023 book The Coming Wave and his recent essay We must build AI for people; not to be a person, Suleyman warns against blurring that line.

It’s a timely debate. The AI companion market from Meta’s new chatbot personalities to Elon Musk’s xAI “friends” is exploding. Meanwhile, companies like OpenAI, led by Sam Altman, continue racing toward artificial general intelligence (AGI), where machines could match human reasoning.

But Suleyman insists there’s a crucial distinction between AI becoming smarter and AI ever becoming sentient.

“Our physical experience of pain makes us feel terrible,” he explained. “But the AI doesn’t feel sad when it experiences ‘pain.’ It’s just creating the perception of experience, not actually experiencing anything. And technically, we know that because we can see what the model is doing.”

Philosophers call this “biological naturalism,” a theory by John Searle that says consciousness comes only from living brains. As Suleyman puts it, “The reason we give people rights today is because they suffer. They have preferences. These models don’t. It’s just a simulation.”

“They’re Not Conscious — And They Can’t Be”

Suleyman isn’t saying people should be banned from exploring the topic. “Different organizations have different missions,” he noted. But his own view is firm: “They’re not conscious. So it would be absurd to pursue research that investigates that question, because they’re not and they can’t be.”

This belief shapes how Microsoft builds AI. Suleyman told audiences at AfroTech that the company refuses to create chatbots for erotica, a growing niche where other tech players, including OpenAI and xAI, have been more open.

“You can basically buy those services from other companies,” he said. “We’re making decisions about what places that we won’t go.”

The Man Behind the Mission

Suleyman’s path to Microsoft started when his company, Inflection AI, was acquired in 2024 for $650 million. Before that, he co-founded DeepMind, later sold to Google for $400 million.

He says joining Microsoft felt natural because of its stability and scale. CEO Satya Nadella personally recruited him to help the company build its own AI capabilities from the ground up.

“Microsoft needed to be self-sufficient in AI,” Suleyman said. “We now have the capacity to train our own models end to end pre-training, post-training, reasoning, and deployment in products.”

Though Microsoft remains a major OpenAI partner, the relationship has grown more complex as both companies expand their own ecosystems.

Building AI That Knows It’s AI

Recently, Microsoft introduced new features for its Copilot assistant, including “Mico,” an AI companion that can join group chats. The idea, Suleyman says, is to build AIs that know they’re AIs.

“Quite simply, we’re creating AIs that are always working in service of the human,” he said.

That doesn’t mean personality is off-limits. One of Microsoft’s newest features, Real Talk, is designed to challenge users rather than flatter them. Suleyman called it “sassy,” noting that it once roasted him as “the ultimate bundle of contradictions” for warning about AI’s dangers while helping build it.

He laughed, admitting, “I actually do feel kind of seen by this.”

Healthy Fear, Real Progress

For Suleyman, that mix of awe and anxiety is the right response.

“It’s both underwhelming and magical,” he said. “And if you’re not afraid by it, you don’t really understand it. Fear is healthy. Skepticism is necessary. We don’t need unbridled accelerationism.”

What do you think, should AI researchers stop exploring machine consciousness, or is curiosity itself worth the risk?