strong AI
Human-equivalent artificial intelligence—machine intelligence with the general capacities of a human mind.
As defined in 17 | Annalee Newitz on Science, Fiction, Economics, and Neurosis
Other definitions
Artificial intelligence systems that are not only smarter than humans in specific domains but achieve general-purpose superintelligence equivalent to or exceeding human capability across all domains.
Reid Hoffman & Peter Thiel The Future of Revolutions
Where it comes up · 4
- 17 | Annalee Newitz on Science, Fiction, Economics, and Neurosis3 claims
- 53 - The Dawn Of Artificial Intelligence2 claims
- Reid Hoffman & Peter Thiel The Future of Revolutions1 claim
- Don’t Believe AI Hype, This is Where it’s Actually Headed | Oxford’s Michael Wooldridge | AI History1 claim
Most important claims · 7 of 7Search all claims for “strong AI” →
- The distinction between strong AI (machines that really understand and experience intelligence like humans) and weak AI (machines that simulate understanding) is philosophically interesting but not practically relevant to most AI researchers, who focus on pragmatic capabilities rather than consciousness.Don’t Believe AI Hype, This is Where it’s Actually Headed | Oxford’s Michael Wooldridge | AI History
- Nothing in the universe has moral value unless it has conscious experience; complex physical processes like exploding stars or rivers have no moral value, so the things that generate moral value are those with conscious experience—and AI as a field has nothing to say about consciousness.53 - The Dawn Of Artificial Intelligence
- We will have a hundred different kinds of artificial intelligence before we have anything truly like a human being, because human-likeness is far harder than commonly assumed—it requires not just processing power but embodied motivations from a body that gets hungry, thirsty, and wants to reproduce—so the first several generations of AI will be nothing like humans.17 | Annalee Newitz on Science, Fiction, Economics, and Neurosis
- As machine learning advances, we are realizing we do not actually know what intelligence is in humans or anything—we are at square one—and trying to recreate something we cannot even quantify in ourselves is extremely hard, which undermines confident attempts to build human-equivalent AI.17 | Annalee Newitz on Science, Fiction, Economics, and Neurosis
- Real human-level artificial intelligence would have to be self-modifying; conversations about how to program a computer so it can't do bad things miss the point, because something that cannot change its own mind in a profound way would not qualify as intelligence—it would at best be equivalent to a horribly abused, culturally programmed human who was never allowed to make choices.17 | Annalee Newitz on Science, Fiction, Economics, and Neurosis
- History shows technology has replaced lost jobs with new ones over 200 years, but predicting whether AI/robots would follow this pattern is premature; if strong AI and humanoid robots smarter than humans and requiring no payment emerged, that would be genuinely scary for middle class employmentReid Hoffman & Peter Thiel The Future of Revolutions
- Strong AI, a term from John Searle, originally meant that building human-level intelligence inevitably produces a conscious device because the functional properties of intelligence and consciousness are inseparable, while weak AI builds systems with desired capabilities without necessarily having consciousness; modern usage shifts 'strong'/'general' AI to mean human-or-greater capabilities with no claim about consciousness, and 'narrow' AI to mean systems lacking generality.53 - The Dawn Of Artificial Intelligence