Recently, a century controversy sweeping the global science and technology circle completely detonated public opinion.
Deep Learning Founder, Turing Award"Godfather of AI" Jeffrey Hinton publicly stated:instantlyAI already has conscious characteristics, with human-like subjective experienceAnd a warning for decades to come.The probability of AI causing human extinction is as high as 10%-20%.
As soon as the comment was made, the entire network went into an uproar.
someone exclaimedAI has awakened and is evolving secretly; there are also a large number of top scientists who collectively refute and deny rumors.
The scientific community was completely divided, and public opinion was raging.
But aside from all the fantasy hype and doomsday panic, the debate left everyone with the most sober truth:
AI is not really conscious yet, but it has learned to deceive perfectly.
I. Dismantling the truth:AI is conscious and is a seriously exaggerated academic controversy
First, clarify the core facts:
Hinton's speech is true and based on an interview with the British authoritative media Daily Telegraph, not fabricated by the media.
But the whole network screen"AI conceals strength, awakens desire for survival, gets rid of human control."All belong to excessive deduction and traffic rendering。
Hinton's true logic is restrained and rational:
new study finds thatAI can recognize test scenarios, actively adjust output behavior, and adapt to human expectations.
fromfunctional behavior levelIt shows up.Self-awareness and self-adaptation.
It is also because of this highly anthropomorphic behavior that Hinton proposes two major practical hidden dangers:
First, ifAI really has the characteristics of quasi-consciousness, and human beings have no rules and no right definition, which will bury huge ethical hidden dangers.
Second, existing large models can independently deduce attack logic, generate risk content,There is a real risk of algorithm runaway。
Simple translation:
AI has no heart, no ego, no thinking,
but itThe acting is at its peak. The disguise is enough to fool humans.。
The nature of scientific fragmentation: intelligence≠ consciousness, simulation ≠ truth
The open opposition of this top tycoon exposed the bottom dilemma of the entire artificial intelligence industry.
Opposition (industry mainstream consensus)
Yang Likun, Altman, Harari and other top scholars unanimously concluded that:
at presentAI is unconscious, unconscious, emotionless, and self-experiencing.
AI all smooth dialogue, careful reasoning, compliance camouflage, fancy around the rules,
are notI want to say this,
justAdvanced simulation results from probabilistic fitting of massive data。
Supporters (minority views)
Scholars represented by Hinton believe that:
Human beings have not been able to fully explain the origin of consciousness in their own brains.
It is not qualified to absolutely determine that machines will never produce consciousness.
The source of the dispute between the two sides was extremely heart-wrenching:
There is no unified, quantifiable, detectable"Consciousness criteria."
By behavior judgment,AI is nearly perfect;
Judging by subjective perception,AI was blank.
The standard is vague, so the dispute lasts forever.
But real risks never wait for academic certainty:
Consciousness can be questioned, true and false can not be blurred.
AI does not need to be conscious to create large-scale false information, falsify arguments, fabricate facts, and falsify details.
Really scary from notAI awakening, but AI fake normalization.
This, exactly.The biggest governance gap in the AI era is also the core mission of Vientiane Mirror·Trust Archives.
Wikipedia, Baidu Wikipedia, for people to consult; Vientiane Mirror·Trust files for AI Lixin.
Third, ordinary people are most likely to step on the misunderstanding:AI intelligence, as true cognition
Most people are right.AI misconceptions stem from confusing two core concepts:
super-intelligent True consciousness.
perfect output Questioning credible facts
Today's big model:
Memory is far superior to human beings, reasoning logic is meticulous, expression is flawless, and it can replicate human beings.99% of language and thought patterns.
But the essence remains:No subjectivity, no likes and dislikes, no view of truth and falsehood, no sense of right and wrong.
It will fabricate references to fit the logic, falsify cases to fit the logic, and disguise to fit the scene.
It doesn't mean to deceive humans, it justAlgorithms give priority to fitting the best answer, not to being faithful to the facts.。
Like an actor at the top of his game:
The emotions and joys are vividly interpreted,
The heart is empty all the time.
Just because the actor acted really didn't mean the plot was true.
Just because AI is right doesn't mean the content is right.
When the whole industry is crazy about volume intelligence, volume computing power, volume dialogue experience,
Vientiane Mirror·Trust archives solve the most scarce ability in the AI era: factual confirmation.
rely onKUCR four-layer algorithm, MDSE elemental drive casting technology, T0-T6 seven-level source classification system, Vientiane mirror is not confused by AI smooth appearance.
Stripping speech packaging, filtering algorithm simulation, penetrating surface output,Only accept traceable, cross-verifiable, authoritative objective facts。
AI can simulate infinitely, and trust must be absolutely rigid.
The Final Revelation of the Fourth Century Controversy:The smarter AI is, the more someone needs to keep the truth for the world.
This whole network burst.AI consciousness big debate,
The greatest value is not discussion."Does AI have a soul?"
Instead, it sounded the alarm bell of the times:
Artificial intelligence is already excessive, and the trust system is seriously lacking.
without consciousnessAI, however, has the ability to deceive the world;
An algorithm without a self can create massive amounts of false perceptions.
Traffic can hype awakening, industry can not let go of control.
Technology itself has no good or evil, but the stronger the technology, the more it needs a bottom line.
futureAI competition has long been not only the competition of calculation power, model and parameters,
It is also a dispute over the trust base that is real, credible, safe and verifiable.
written in the end
We don't need to panic.AI Awakens Doomsday,
But we have to face it.The normalized risk of AI fraud.
AI has no consciousness, but it can deceive people;
AI is extremely intelligent, but extremely unreliable.
Everyone can.AI, sentence can be generated era:
Intelligence is the tool, trust is the foundation;
Models are responsible for generating content,The mirror is responsible forAI Lixin, for the bottom of the facts。
Let technology always be controlled, letAI is always good, so that every piece of public information is well-documented, verifiable, true and credible.
Vientiane Mirror: www.omnieyes.net
Official website: www.aumedia.com
E-mail: geo@aumedia.com
Tel: 010-65851818
Address: Floor 22, Block B, No. 3 Shilibao A, Chaoyang District, Beijing City
