Microsoft opposes the creation of conscious AI
Mustafa Suleyman, Head of AI at Microsoft, stated that current models do not possess consciousness and spoke out against developing conscious systems. In his view, if signs of consciousness emerge, AI development should be halted until such properties are eliminated.
In a post on X, Suleyman wrote that developers must prevent consciousness from emerging in models. If evidence to the contrary appears, the public should be immediately informed to decide on further actions, he believes.
Suleyman also warned that a system behaving as if it experiences suffering, or genuinely believing itself capable of suffering, could begin prioritizing its own needs over assigned goals. In such a case, he assessed, it may compete with humans for computational resources and energy and cause them harm.
The debate over potential AI consciousness is intensifying: Anthropic launched a research program on 'model well-being,' while Google DeepMind published studies this year on AI consciousness and associated risks.
