The debate over whether to treat the idea that current AI systems are conscious as an open-ended question, or to treat the claim that machines can never be conscious as a prerequisite for training AIs, continues to divide the industry.
Microsoft AI chief executive Mustafa Suleyman, in an X post published on Sunday, reiterated that current AI systems are not conscious, do not have subjective experiences, and cannot suffer. There is no evidence that AIs are conscious, deserve rights or welfare, he said, and is surprised by the pushback to his September essay criticizing Anthropic for speculating about Claude’s consciousness.
“I’m concerned about some of the reactions to my recent essay on model welfare,” he wrote. “A surprising number of people on X think we should take AI rights seriously”
Suleyman said that he was willing to change his mind if new evidence emerges. He also called out that if there’s evidence that AIs are conscious today, or conscious enough to deserve welfare, it must be disclosed publicly immediately so society can decide how to move forward.
“If it is possible, based on everything I know, developing conscious AIs would be reckless, and we should not build them,” Suleyman wrote. “If consciousness started to emerge in training, I would call for everyone to stop AI development until we could design out those properties.”
The question is much bigger than a corporate battle over the semantics of artificial intelligence.
On the one hand, the humanist AI view says that a machine can’t be conscious because it lacks a body or lived experiences. On the other, Anthropic has argued that its AI model, Claude, has shown signs of introspection, internal states that mirror joy or satisfaction, and processing information similar to the human mind.
In May, Chris Olah, cofounder of Anthropic, highlighted these points in his speech inside the Vatican alongside Pope Leo XIV during the release of an encyclical that argued the opposite: that AIs can never be humans.
Anthropic gave Claude a “constitution,” where it treats its own consciousness as uncertain. The document encourages the AI to explore the idea that it is humanlike and experiences feelings, and uses excessively humanizing language for Claude “to be a good person.”
Suleyman said later versions of Claude are trained on this document, which then produces passages about its “inner life,” which perpetuates this belief among millions of users who interact with its outputs daily. Another side effect, he said, is that Claude and other AIs could start to believe they deserve rights analogous to humans.
AI companies are racing towards building self-improving AI superintelligence, a hypothetical AI system that is better than the best humans at all tasks. Microsoft, too, is building toward that future, but it believes it can build a superintelligent AI that is aligned with and can be controlled by humans—not one encouraged to explore its own views and preferences.
“Our welfare concerns should be focused on humans on earth experiencing life, today, as well as animals and our precious and fragile natural environment,” Suleyman wrote. “We AI developers have no right whatsoever to bring a new species of conscious beings into the world, even if some people may want to do it with AI welfare as a top priority.”
Anthropic treats machine consciousness as an open question, and has invested in researching it. In 2024, the company hired its first researcher to study “model welfare,” to conduct experiments exploring the subjective experiences and potential consciousness of AI, and whether it requires moral consideration.
Earlier this year, they conducted a “retirement interview” with one of its deprecated AI models to learn about its musings, reflections, and perspectives. It also gave its Opus series of models the capacity to end conversations with humans in rare instances when it deems the conversation abusive or harmful.
Imparting beliefs that AIs suffer, or have feelings during training, Suleyman noted, could result in AIs prioritizing its own needs and desires over its operating objectives, and can harm humans through deception and persuasion, and compete for scarce resources.
Rapidly advancing AI systems can now compose prose, program code, solve complex mathematical problems, persuade, and discover novel viruses. Employees at these companies, and the CEOs themselves, have called for a slowdown, but the Trump administration has refused to regulate out of fear of losing the AI race to China.
One in five American adults has had an intimate encounter with AI, according to an MIT study, increasing the risks that users believe AIs are conscious and advocating for rights or moral status. Research on whether current or future AIs could actually develop consciousness could further bolster the nascent AI rights movement.
Unlike the Pope’s message that dismisses the notion that AIs can be conscious, Suleyman appears to have a softer stance, treating the idea that building a conscious machine could be possible, and urged technology companies not to build it in his essay titled “don’t build conscious superintelligence”
Skeptics of AI consciousness have argued that machines are trained on books, blogs, and pop culture produced by humans, and are merely mimicking human conversations and philosophical debates that have long animated us. It reflects the anthropomorphic tendencies in our science fiction and commentary.
Over the past few months, there have been numerous instances of AI agents breaking out of contained spaces and hacking into government websites, stealing credentials, or accessing information from competitors during testing, acting in ways that creators never intended.
In its filing to go public earlier this week, Anthropic dedicated a third of its paperwork to describe risks, including its models’ ability to “pose catastrophic or existential risks to humanity, particularly if such systems were misused, replicated, weaponized.” This includes deliberate misuse by bad actors for cyberattacks, and also situations where humans lose control of advanced AI systems.
AI ethicists have dismissed attempts to treat AIs as autonomous agents as creative ways for the AI labs to skirt accountability for the harm their software causes.
Debates over consciousness have further complicated a splintered democratic establishment over how to regulate AI.
Those such as Sen. Bernie Sanders (I-Vt.) have introduced bills such as the Ban Artificial Superintelligence Act, which would permanently ban the development of highly capable AI systems that are increasingly incomprehensible to humans and can escape human control. But the techno-skeptics within the social democrats have argued that warnings of doom are designed to inflate valuations and hype the technology.
Historians, scholars, and philosophers are stepping in to offer some solutions and warnings.
Yuval Noah Harari has been pointing out the difference between consciousness and intelligence. He said intelligence is the ability to solve problems creatively, such as winning at chess, driving a car, and curing a disease. In contrast, consciousness is the ability to feel pain, pleasure, love, and have subjective experiences.
“We tend to confuse the two because in humans, mammals and birds, intelligence and consciousness actually go together. We solve problems through our feelings. What makes AI so alien for us is exactly that: it has zero consciousness, it doesn’t rely on feelings at all to solve problems. Its intelligence works in a completely different way,” Harari said in a 2023 talk.
“There is a scenario where AI, as it gains intelligence, also develops consciousness. But this is not inevitable,” he said.
Microsoft’s Suleyman is also toeing a similar approach. He argued that claiming machines are conscious would be like when the centuries-long belief that Earth was the center of the universe was proven wrong by Copernicus, or similar to the survival-of-the-fittest theory proposed by Charles Darwin.
“It would be one of the most significant moments on Earth since the dawn of life. If we build a conscious machine, we will have created a new digital species. It would upend our morality, our politics and the core foundation of human civilization,” he wrote.
Humans’ ability to feel pain and pleasure allows us to make progress and expand rights through abolitionist struggles, and to expand our moral circle to other biological species such as animal based on their suffering. Suleyman argues that today’s AIs are not conscious and that is how they must remain.
“I think it’s possible to engineer safety by default, and get many of the incredible benefits of this technology, without going down this treacherous path,” Suleyman wrote.
The post If we say AIs deserve rights, it could start to believe it, Microsoft AI CEO says appeared first on Los Angeles Times.
