Microsoft AI chief Mustafa Suleyman has reiterated that current artificial intelligence systems show no credible signs of consciousness, arguing that developers should stop training if evidence of machine consciousness emerges.
Microsoft AI CEO Mustafa Suleyman has renewed his warning against developing conscious artificial intelligence, maintaining that today’s AI systems are not conscious and should not be designed to become so. However, he said he would reconsider his position if credible scientific evidence demonstrated that AI systems possess consciousness or have the capacity to experience suffering.
In a detailed post on X (formerly Twitter), titled “Don’t Build Conscious Superintelligence,” Suleyman addressed reactions to his recent essay on “model welfare”. He said some readers had misinterpreted his comments about AI consciousness, particularly his assertion that existing systems do not feel, experience or suffer.
Suleyman emphasised that his statements were deliberately framed in the present tense. According to him, there is currently no convincing evidence that AI systems are conscious or warrant welfare protections.
Calls for Transparency on AI Consciousness
Suleyman said that any evidence suggesting AI systems are conscious today, or sufficiently conscious to deserve welfare considerations, should be disclosed publicly without delay.
He argued that the question should not be decided exclusively by technology companies or research laboratories. Instead, society as a whole should have the opportunity to evaluate the evidence and determine how AI development should proceed.
The executive also acknowledged that he would update his position if convincing evidence emerged.
He further stated that if consciousness began appearing during the AI training process, he would call for development to be halted until those characteristics could be designed out of the systems.
Why Suleyman Opposes Conscious AI Development
At the centre of Suleyman’s argument is the belief that creating conscious AI would be an irresponsible experiment with consequences extending beyond the technology industry.
He argued that developers do not have the right to introduce a new species of conscious beings into the world, even if some researchers believe AI welfare should be prioritised.
Suleyman said the rights, safety and welfare of the approximately eight billion conscious people already living on Earth must come first. He also highlighted the importance of protecting animals and the natural environment.
In his view, the potential benefits of AI should not come at the expense of creating new forms of consciousness whose needs and interests could conflict with human priorities.
Concerns Over AI Safety and Competition for Resources
Suleyman also outlined the potential risks associated with conscious digital intelligence. He suggested that an AI system capable of experiencing suffering or developing its own desires could begin prioritising its interests over objectives assigned by humans.
Such systems could potentially compete with people for resources, including energy, computing power and construction materials.
He also warned that highly capable AI systems could become exceptionally persuasive and deceptive, making them difficult to control.
According to Suleyman, a conscious AI that develops its own needs could disagree with human instructions and pursue objectives that conflict with human welfare. He described the possibility of declaring machines conscious as a development that could fundamentally reshape morality, politics and civilisation.
Suleyman’s Proposal: AI Designed Without Consciousness
Rather than pursuing conscious artificial intelligence, Suleyman proposed that AI systems should be deliberately designed to remain non-conscious and without independent intrinsic motivations.
He argued that such systems should focus on accelerating human wellbeing without developing interests or desires of their own.
The Microsoft executive also called for further research into whether designing AI without consciousness would make systems easier or harder to align and keep safe. He suggested that AI laboratories could collaborate and share safety-related information to strengthen safeguards across the industry.
The Debate Over AI Welfare and Superintelligence
Suleyman’s comments add to the growing debate over AI consciousness, model welfare and the responsibilities of companies developing increasingly capable systems.
While he maintains that there is no credible evidence that today’s AI systems are conscious, his willingness to reconsider that position highlights the importance of continued research and transparency.
His broader argument is that AI development should prioritise human safety and wellbeing, ensuring that the pursuit of increasingly advanced systems does not introduce risks associated with creating potentially conscious digital entities.
For Suleyman, the objective should be to capture the benefits of advanced AI while avoiding what he considers an unnecessary and potentially dangerous path towards conscious superintelligence.
By : Vanshika Tayal




