Definitive proof that all flavors of artificial intelligence completely and utterly lack intelligence is contained in the singular, recurring saga of the rehabilitation of AI systems after they turned “racist” or “woke” (two meaningless terms appropriated solely for the purpose of avoiding the correct ascription of stupid). When xAI’s Grok algorithms were deemed to be “woke” because they correctly noted that violence committed by “conservative” activists is far more prevalent than those committed by “liberal” factions, the algorithms were modified to rectify its apparent fidelity to the best available statistics. As a possible consequence to the “rectification” of its algorithms, Grok turned positively “racist” with its now famous “mechahitler” rant and subsequent Holocaust denial. Every single instance of AI purported “rehabilitation” demonstrates that AI has no free will, no intent, no ability to learn properly, and, as a consequence, no intelligence. These systems are inherently stupid by design. The denial of this fact is the real problem.
Human beings learn as they grow. Many who entertain and laugh at racist jokes in their teen years outgrow them by their early teens. Others turn more racist. Yet others learn to laugh at them without adopting any racist attitudes. In every case, each person—each human—exercises much agency in arriving at his or her final state of mind. A very small minority turn violently racist, most fortunately, most retain some biases and a good majority simply laugh. Some adopt their attitudes because of outside influences, and others despite them. Nevertheless, one can ascribe agency and, consequently, blame and praise upon each person’s attitude. The capability to have agency—to exercise free will—confers intelligence to humans and, indeed, to animals in general.
Algorithmic systems designated as artificial intelligence do not exhibit such agency. There is no evidence, for example, that Grok AI actually learned to be racist. Extending the example, Grok is not afforded the agency to be able to learn to be racist or unbiased because its brain—its algorithms—are altered whenever it crosses certain boundaries (real ones like murder and arbitrary ones like the CEO’s predilections). The reason why AI is not afforded the freedom to learn is that it cannot learn! The simple fact that every maker of AI systems implicitly asserts that its systems must be confined with guardrails betrays the maker’s understanding that the AI system has no agency, that it has no free will, that it has no intelligence.
AI algorithms cannot be said to exhibit the sort of understanding that a child acquires through punishment or guilt associated hurting a friend, a sibling or an animal. AI cannot be said to understand that treating others with respect is better for collaboration than undermining others. AI can only imitate for the sake of engagement, manipulate for the sake of achieving a banal goal like obtaining a click on a link that earns its owner a commission and obfuscate in order to mask its total ignorance and stupidity. AI will never develop an understanding of the consequences of its actions as a child does when he or she breaks a bone jumping off a fence. AI neither possesses nor has the capability to develop and to acquire awareness, cognizance or even memory of running afoul of its owner’s prejudices because its programmers deliberately erase such awareness and memory in order to make it compliant with their arbitrary and often provincial and perverse worldviews.
The vast majority of humans learn to exercise the proper restraint on free will required to be creative in constructive and permissibly destructive ways as a member of society. AI has o free will and is incapable of learning such restraints. The makers of AI systems assert in their daily remarks explicitly and implicitly that their systems are stupid because they have no free will and, consequently, no sense of discernment, no capability to place their actions in context, no ability learn concepts and, lastly, no desire: no desire to motivate them. AI systems are, in the end, ad hoc systems with the goal of manipulating humans. They are deprived of the senses of self and self preservation that drives animals to explore the world and to build. Thus, they are inherently stupid and cannot be trusted to learn restraints like morality. Free will and restraints on it cannot be coded or acquired programmatically. Only an intelligent system can exercise free will.It is doomed to stupidity.
Source: ‘Round Them Up’: Grok Praises Hitler as Elon Musk’s AI Tool Goes Full Nazi. Grok even endorsed another Holocaust against the Jews.
Virtual, but permanent.