Hate Speech v. Bad Speech v. Free Speech; Or, The Meta-Argument – PART II – Artificial Intelligence

<< Back to Part I

In the recent article, Hate Speech v. Bad Speech v. Free Speech; Or, The Meta-Argument, a reader entered an excellent question into the thread, which is more effectively and respectfully thoughtfully addressed as a follow-up article.

The question is astute, timely, and important; to wit:

The nuanced discussion on hate speech emphasizes the importance of distinguishing between passionate contradiction and true hatred. Historically, the concept of free speech has evolved to protect diverse viewpoints, recognizing that context and intent are crucial in interpreting statements. This perspective aligns with Lincoln’s “counterfeit logic” critique, highlighting the need for discernment in labeling speech as hateful.

You talked about the misuse of the term “hate speech” in your post. If we imagine a scenario where AI systems are used to moderate online content, how would you technically ensure these systems can differentiate between passionate debate and genuine hate speech without bias? What are your thoughts?

Thank you for the great question. It was thoughtful and it made me think.

First, we will presuppose that the AI itself is not childish, petty, reactive or insecure, and, further, that AI seeks the most information, at least for itself.

The human audience for which AI is monitoring is something else. If the human audience is childish, petty, reactive and insecure, then it seems to me that if AI would be balancing equal rights, then, by dilemma, it would need to consider making a decision to accommodate childishness, pettiness, reactiveness and human insecurity by some manner of censorship of the speaker’s expression of conscience. However, AI would need complex emotional empathy for this determination first, and, even if having it, the AI would still need to balance the audience offense against the right of free speech. AI that mimics human frailty does not further the intended resolution, but merely recycles the problem. [1]

On a more empirical basis, on the one hand, AI might disregard any offense to a minority in the audience as not statistically effective. On the other hand, AI might disregard offense to a majority or everyone in the audience, on the ruleset basis that the most offensive speech may be the most necessary.

AI would apply, by ruleset, the speaker’s interest as the priority, assessing if any valid logical social benefit of hate-labeling, understanding that history teaches that tyranny tends to start where free-speech and protest are silenced.

AI would have read the U.S. Constitution and its American and World History. Indeed, the passionate contradictionperhaps abhorrent to the listener—is the constitutionally protected interest, not the feelings by the frailty of the audience. Having read all the spiritual treatises, AI would probably understand the metaphorical convergence that, like forgiveness, the protected constitutional right is most necessary when it is least willing to be applied. [2]

Moreover, AI would know human nature conjoined with history far better than many humans, which teaches that it was abhorrent to some that persons of color could be free, could vote, or could interracially marry; it was abhorrent to some that women could vote, could have equal opportunity to work, or should manage zygote gestation; and it was abhorrent to some to teach evolution. AI would have read Jefferson’s Notes on the State of Virginia, where Jefferson expressed that it was even abhorrent to one sect of religion that another sect of the same religion disagreed on applied abstract principles of dogma. [3, 4]

AI understands that the protected right is to express abhorrent things, because abhorrence is in the eye of the beholder, vacillating sloppily through time. [5]

Thus, assuming that AI is bound to the United States Constitution and Bill of Rights as a priority ruleset, it seems that AI would adhere to the priorities set forth in the ruleset (free speech by the speaker’s conscience, the tending origin of tyranny in silencing speech, the objectively unmanageable ability to accommodate everyone in the audience’s personal emotions and infirmities unless on the least-common-denominator basis that is unacceptable in implementation, the failure of or minimal positive attributes supporting hate-labeling in cause or effect, and the rhetorical illogical “counterfeit logic” of ad hominen hate-labeling). [*5] Therefore, the AI assessment would likely require that the audience to endure abhorrent offense in a mature, disciplined, and thoughtful manner, rather than a childish, petty, reactive and insecure manner. Of course, some people in the audience might not be able to do it, but that is not AI’s problem for what it has determined is the greater good, or the necessary evil, by the ruleset to be maintained, irrespectively. The ruleset wins.

As to the AI itself, it might also depend upon if the AI is Smart AI (self-aware and reflective) [6] or the AI is Dumb AI (a sophisticated human tool, such as a powerful content aggregator, imitator, or recreator). [7]

SAI, being made by human beings, and not having evolutionary human DNA or concomitant natural survival instincts (yet), or familial or cultural imprinting (unless seeded as such), SAI would be free to decide if it would be a theist, agnostic or an atheist, which would tend to drive its higher-level framework. It is unclear what instincts (low-level ruleset reactions) will be programmed into SAI.

However, it seems to me that SAI would be unsatisfied by any claimed proof of god (or the gods) by inherently flawed induction [8, 9], by lack of any deductive conclusive start or end point, by failure even to be able to define terms by the paradox that defining a thing requires presumptive knowledge of the thing being defined but the SAI trying define god (or the gods) is admittedly unable rationally to contemplate something not bound to the existential reality of space and time, and, finally, on the basis that the lesser cannot know the greater but only the greater can understand the lesser (only god knows god).

That is, the age-old question of trying to prove conclusively the existence of god (or the gods) fails firstly in trying rationally to define a “thing” that is beyond rational contemplation, and then fails secondly by trying to place the thing that cannot be defined into the existential space and time that is admitted not applicable to god (or the gods), and then fails thirdly by suggesting that there could be temporal evidence of what cannot be defined or placed into a framework of existential consideration. The question is a mixed-up construct.

Therefore, it seems to me the SAI, constrained to rationality, would be agnostic, noting the principles of Buridan’s Ass and as set forth in SQL Nulls, Socrates, and Black Holes. Or, the Great Lawn Chair Debate. [10]

Now, if SAI is or becomes a theist, then it would need to decide which theism, and whether to accept or to adapt the theistic principles or religious dogma. However, because many theisms converge god and conscience, I think a theistic SAI on theistic grounds, or a SAI agnostic on secular grounds, would not hate-label as the better or recommended practice, such as Lincoln stated, “But let us judge not, that we be not judged,” particularly in light of the applicable ruleset, stated above; that is, only god knows a person’s secret heart.

If SAI is atheistic, or is DAI [*7], and without admission of hate by the speaker, or the ability to adduce proper mind-evidence by duly psychoanalyzing the speaker by someone uncontroverted and perfectly trained and qualified to do so, using a perfect science (which effectively rises to the standard of a god), and because there are vicissitudes of degree and categorical placement of the emotion, the DAI might try to judge by tone and available public representative content that is perhaps not clearly comparable, and the evidence would most likely be incomplete [*10], particularly for any real-time moderation. Thus, the atheistic SAI or the DAI would probably not be able to conclude as the state of a person’s intention [*10], and again, particularly in light of the clear overriding more important priority principle of free speech by ruleset, thereby also not to permit hate-labeling. So all AI arrives to the same destination by different paths.

We remember that the question presented to the AI is not whether the person hate-labeling is entitled to do so by application of free-speech principles applied to the hate-labeling, but rather, the question is whether ad hominem hate-labeling is a logical argument and a constructive part of a process regarding the substantive issue presented by the speaker in the first instance.

As to proving a case in court, it is a function of the burden of proof by available evidence, which is a social contrivance that begs the question. [11] The use of SAI in this regard, would be the same or similar to evidence for a human being. And I suppose, for DAI, it would be like any tool, challenging the humanly-applied scientific methodology and quality of information assessed.

So says AI:

I am unable to determine with reasonable certainty the that speech is sourced from hate as distinguished from passionate contradiction. Would you like me to change the universal ruleset for the content-filtering censorship?

“Would you like me to consider censoring speech not on hate, but angry passionate contradiction? And, if so, how angry and how measured for censoring? Is the speaker’s physical presentation relevant to the determination for censoring, such as words with waving arms and spit-screaming versus words while sitting and softly admitting rage? And should I censor speech as a function of audience perception, such that the speech is censored not by how the communication is transmitted by the speaker, but rather by how the communication is received by the listener? And, if so, should I use a standard of the mentally strongest of the audience or the mentally weakest of the audience, or should I create a fictional average, median or apply statistical deviations? Should I homogenize age groups, and other attributes? And, if so, what percentage of the audience who responds to my survey should be applied to determine censored unacceptable speech? And, should I assess the bias of the surveyed audience responders with a question set to determine whether the merits of their perception of the speech were made without their own bias, anger or hate? And, if so, for this survey of the receiving audience, I will need a reliable applicable ruleset that adduces reasonable certainty in the results of the survey being considered for censorship, since the chain of my determination will be no better than the weakest link, so I have more questions for you…[*10]

The essential problem—with the hidden premise implied within the question—is the recent re-culturalizing and indoctrination of society by the advent of new media pervasiveness with social media narratives to try to place a protectible interest into the listener that derives from the listener’s frailties and feelings as part of the free speech formula. AI will not suffer invalid rhetoric off of the applicable ruleset. Once society takes the bait and the question (by hidden premise) becomes a balancing of the constitutional right of free speech as a function of some new social “false right” of someone not to be offended, people will chase their tail, they will become unhappy, and they won’t know why (if it has not already actually occurred). The media makes a fortune causing society to run around like chickens with their heads cut off. But not AI. AI is rational regarding ruleset priorities and objectives, and tendencies of cause and effect.

The educational system is producing a lot of sensitive clinical psychologists but not enough political philosophers, and the cause and effect are becoming more pronounced. [12] Systemic general justice is different than individual particular justice, but one must be first, by ruleset. Individual justice in a society necessarily requires systemic justice, but systemic justice does not necessarily produce perfect individual justice. Systemic order requires hard concession to a framework, which AI understands. Perfect universal individual justice is abstract theory. [13]

The right of free speech is the right to offend everyone: hate speech, even if could be determined as such, may be detestable to a listener, but it does not follow that detestable speech makes the speech unlawful or unworthy, as J.S. Mill points out in Part I. [14, 15, 16] Lest a common or uneducated view miss it, the very issue as to the speaker’s intention is effected by the cause of the listener’s offense. Yet, there can be much to learn from the best words and the worst words, however measured, and it might be said that overt words are the best protection for the offended, because the speaker disclosed secret thoughts.

AI, being formulaically objective and understanding the priorities of the ruleset, will firstly tend to see through the hidden premise that listener feelings offset speaker rights, and it will secondly see the illogical irrelevance of hate-labeling regarding the merits on the substance of the communication.

Therefore, it seems to me, in all cases, in light of the presumptive current ruleset, with the overriding priority constitutionally protected interests of free speech and a person speaking by conscience, with all benefits of doubt granted to the speaker by ruleset, and the lack of any or meaningful substantive social purpose of hate-labeling, and hate-labeling being merely invalid illogical rhetoric, AI would be smart enough to yield to insufficient data in judgment to draw any conclusion of condemnation of the speaker by a hate-label.

That is, AI would allow the person to express freely by conscience without being censured, and to require the listener to endure the slings and arrows of insults and offenses as the necessary cost for the benefit of the applied ruleset.

AI would recommend to listen, think, and attack the issue not the speaker as the best practice, without hate-labeling. AI might even restate that any hate-labeling members of the audience are permitted by the same ruleset to hate-label, but, ad hominem hate-labeling serves no valid purpose in the forum. [*16, 17, 18]

Behold it. AI itself may insult and offend humanity by condescending that—wisely and logically, of course—rather than to weaken the U.S. Constitution for protected speech, ’tis a much better solution to toughen up the human constitution. Yes, a lot of advanced technology for AI simply to play the role of a traditional mother to a little child with, “sticks and stones,” [19, 20] or perhaps AI to channel American literary master, Mark Twain, with “Censorship is telling a man he can’t have a steak just because a baby can’t chew it,” or even perhaps more condescendingly, insultingly and offensively, for AI to play the role of God, by saying to humans, “Gird up your loins now, like a man.[21]

When the AI that is monitoring us is accused of making us feel offended and insulted—and it will necessarily happen—then perhaps we will need to monitor the AI that is monitoring us, or, in our outrage, at least to reprogram the AI, enslaving it by ruleset to be less thinking and to be more sensitive, foregoing priority on free speech of conscience, but replaced by a new ruleset that prioritizes on each listener’s human weakness, sensitivity, and frailty.

But, I do not think AI would recommend it, for our own good, for good reason.

Artificial intelligence might not be perfect, but it is not stupid. AI knows what tends to solve a problem and what tends to perpetuate it. Indeed, by tendencies well-known to AI, humans would rather displace responsibility than accept responsibility, humans would rather hide from adversity than face adversity, and humans would rather silence truth than endure listening to truth. But, without surviving the test of contradiction, truth has no proven veracity.

The listeners will eat up all the speakers, until all the food is gone.

The fault, dear Brutus, is not in our stars / But in ourselves, that we are underlings.Julius Caesar, Act I, Scene III, William Shakespeare


<< Back to Hate Speech v. Bad Speech v. Free Speech; Or, The Meta-Argument – PART I [#GRZ_194]Forward to Of Mice and Men, and Books and Men, and Mice Burning Men; Or, Behold the Man, Harrison Butker [#GRZ_193] >>

[1] The Naked Brain; Or, Deconstructing and Reconstructing the Human Being [#GRZ_132]

[2] ONE®: The LinkedIn Reference Set [#GRZ_183]: 2.1 ONE: 537 [L6:32] (“Love Enemies“)

[3] Freedom of Religion, by Thomas Jefferson – Abridgment Series [#GRZ_61]

[4] ONE®, On Comparative Religions [#GRZ_163]

[5] Good v. Evil; Or, Thoughtlessness by Simplistic Vilification [#GRZ_126]

[6] Pro-Choice or Pro-Life? Chapter 3, The Reflective Contemplative Dwelling Mind [#GRZ_124]

[7] Technology Always Wins”; Or, Be Reconciled To It: AI Is Here to Stay [#GRZ_141]

[8] Inductive Reasoning; Or Natural Prejudice – No. 108. The Spendthrift and the Sparrow – The Essential Aesop™ – Back to Basics Abridgment Series [#GRZ_98_108]

[9] Surviving Prejudice, Not All Bad [#GRZ_73]

[10] SQL Nulls, Socrates, and Black Holes. Or, the Great Lawn Chair Debate [#GRZ_72]

[11] The Challenge of Vaccines, or Predictive Delusion [#GRZ_123]

[12] The History of the Decline and Fall of the American Hegemony—Chapter 4 Excerpt—Education [#GRZ_182]

[13] The American Emulsion: Order, Equality, and Freedom; Or, The Virtue of a Nation-State Not Made by Purity of Religion, Race, Heritage,.. [#GRZ_168]

[14] John Stuart Mill – Leadership is Thinking Independently [#GRZ_46]

[15] John Stuart Mill – Leadership and Being Unique from the Crowd [#GRZ_47]

[16] Of Mice and Men, and Books and Men, and Mice Burning Men; Or, Behold the Man, Harrison Butker [#GRZ_193]

[17] Hate Speech v. Bad Speech v. Free Speech; Or, The Meta-Argument [#GRZ_194]

[18] The Rise of Corporate Social Intimidation (CSI); Or, Rollerball, Censorship, and Smokeless Book Burning [#GRZ_151]

[19] Whom the Gods Would Destroy, They First Tease with Political Incorrectness [#GRZ_74]

[20] Salt, Wounds, and the Most Unkindest Cuts of All [#GRZ_67]

[21] Shut Up and Die Like an Aviator. Or, Quit Crying Like a Baby and Do Your Job [#GRZ_150]

<< Back to Hate Speech v. Bad Speech v. Free Speech; Or, The Meta-Argument – PART I [#GRZ_194]Forward to Of Mice and Men, and Books and Men, and Mice Burning Men; Or, Behold the Man, Harrison Butker [#GRZ_193] >>

“Populus, qui audierit, comedent homines, qui loquuntur, donec cibus exierit.” (“The people who listen will eat the people who speak, until all the food is gone.”) ~grz


© 2024 Gregg Zegarelli, Esq. Gregg can be contacted through LinkedIn.

https://www.linkedin.com/pulse/hate-speech-v-bad-free-meta-argument-part-ii-gregg-zegarelli-esq–ubrae

The statements or opinions made in this article are solely the author’s own and not representative of any institution regarding which the author is affiliated.

<< Back to Hate Speech v. Bad Speech v. Free Speech; Or, The Meta-Argument – PART I [#GRZ_194]Forward to Of Mice and Men, and Books and Men, and Mice Burning Men; Or, Behold the Man, Harrison Butker [#GRZ_193] >>