Author: Michael LaBossiere

 

110 Fallacies

Description:

Like the general Appeal to Silence fallacy, the Gish Gallop and Fire Hose of Falsehoods are tactics that involve taking a failure to respond as evidence for a claim.

As a rhetorical tool, the Gish Gallop is an attempt to overwhelm an opponent by presenting many arguments and claims with no concern for their quality or accuracy. The Gish Gallop was named in 1994 by anthropologist Eugenie Scott who claimed that Duane Gish used this tactic when arguing against evolution.

The Gish Gallop is somewhat like the debating tactic of spreading which involves making arguments as rapidly as possible in the hopes that the opponent will not be able to respond to all of them. The main distinction is that the Gish Gallop is an inherently bad faith technique that relies on rapidly presenting weak arguments, fallacies, partial truths, Straw Men, and lies in the hopes that the opponent will not be able to refute them all. The Gish Gallop can be seen as a metaphorical cluster bomb of fallacies and untruths.

While this technique lacks logical force, it can have considerable psychological force. The Gish Gallop relies on Brandolini’s Law, which is the idea that it takes more time and effort to refute a fallacy or false claim than it takes to make them. Effective use of a Gish Gallop will yield many unrefuted fallacies and false claims, and this can create the impression in the audience that the Gish Galloper has “won” the debate. The Gish Gallop can be combined with Moving the Goal Posts to create the illusion that at least some of the refutations have been addressed.

Psychologically, the side that seems to have made the most unrefuted arguments and claims might appear to be correct, especially if the Gish Galloper uses the Gish Gallop fallacy, which has the following general form:

 

Premise 1: Person A presented N arguments for claim C.

Premise 2: Person B, the opponent, refuted X of A’s arguments.

Premise 3: N is greater than X.

Conclusion: C is true.

 

This is fallacious reasoning because it is not the number of arguments that proves a claim, but the quality of the arguments. As an illustration, consider this silly example:

 

Premise 1: During a debate, Bob presented 123 arguments that 2+2=6.

Premise 2: Bob’s opponent Sally only refuted 2 of Bob’s arguments before time ran out.

Premise 3: 123 is greater than 2.

Conclusion: Therefore, 2+2=6

 

While the error in reasoning is obvious in such absurd cases, people can easily fall victim to this reasoning in more complicated or controversial cases, especially if the audience does not know the subject well.

One reason why this fallacy might be appealing is that it seems analogous to methods that do work. For example, a swarm of relatively weak ants can overwhelm a strong spider in virtue of their numbers, even though the spider might kill many of them. But argumentation usually does not work like that; weak arguments generally do not add together to overcome a single strong argument. So, the analogy is not a swarm of ants beating a spider, but a spider fighting weak ants one at a time.

Another reason the fallacy might seem appealing is that making claims or arguments that are not refuted could seem analogous to one team not being able to block every shot taken by their opponent. But the Gish Gallop would be best compared to a basketball team rapidly taking wild shots all over the place, not caring whether they are even made in the direction on the basket. The opposing team does not need to block those wild shots; they are not going to score any points. In the case of arguments, not refuting a bad argument does not prove that the argument is good. Not refuting a claim does not prove the claim is true. See Burden of Proof for a discussion of this.

While the Gish Gallop technique involves presenting at least some arguments, a related technique is to blast an opponent with a Fire Hose of Falsehoods. In this context, the Fire Hose of Falsehood is a rhetorical technique in which many falsehoods are quickly presented. The technique can also employ the rhetorical technique of repetition. As a matter of psychological force, the more times a person hears a claim, the more likely they are to believe it. But the number of times a claim is repeated is irrelevant to its truth. This method also often involves using multiple channels to distribute the falsehoods. For example, real users or bots on various social media platforms could be employed to spread the falsehood. This can have considerable psychological force since people are also inclined to believe a claim that (appears to) come from multiple sources. But the mere number of sources making a claim is irrelevant to the truth of that claim.

This technique can be used to achieve various ends, such as serving as a Red Herring to distract people from an issue or, in its classic role, as a propaganda technique. On a small scale, such as in a debate, it can be used to overwhelm an opponent because a person can usually tell a lie much faster than someone else can refute it. This technique can be used with Moving the Goal Post to exhaust an opponent and run out the clock.

It can also be employed as a variant of the Appeal to Silence. As a fallacy, the reasoning is that unless all the falsehoods made by someone are refuted, then their unrefuted falsehoods are true. As a fallacy, it has this generally form:

 

Premise 1: Person A makes N falsehoods.

Premise 2: Person B, the opponent, refuted X of A’s falsehoods.

Premise 3: N is greater than X.

Conclusion: The unrefuted falsehoods are true.

 

Laid bare like this, the bad logic is evident. Not refuting a falsehood does not make the falsehood true. When someone uses this fallacy, they will attempt to conceal the logical structure of this reasoning. They might, for example, simply say that their opponent has not refuted their claims and so their opponent must agree with them.

While this is a fallacy, it can be effective psychologically. If a person seems confident in their falsehoods and overwhelms their opponent with the sheer number of their lies, they might appear to have “won” the debate.  

 

Defense: To avoid being taken in by the Gish Gallop, the key is remembering that the support premises provide to a conclusion is based on the quality of the argument. The quantity of (unrefuted) arguments for a claim, by itself, does not serve as evidence for a claim. In the case of claims, a failure to refute all the claims made a person does not prove that the unrefuted claims are true; this applies to both the Gish Gallop and the Fire Hose of Falsehood.

If a Gish Gallop or Fire Hose of Falsehood is being used against you in a debate, you will almost certainly not be able to respond to all the arguments and claims. From a logical standpoint, one good option is to briefly point out your opponent’s technique and why it is defective. If you are arguing for a position, focus on your positive arguments and, if time permits, respond to the most serious objections. If you are arguing against a position, focus on your arguments against that position and, if possible, try to pre-empt the arguments your opponent is likely to use in their Gish Gallop. You can also sometimes group arguments and claims together and refute them in groups. For example, if an opponent uses multiple Straw Men, you can respond to all of these by pointing this out.

 

Example#1

Gus: “So, my opponent is a climate change scientist. That means she hates capitalism, so she is wrong. Also, these so-called climate change scientists say that humans are the only things that affect the climate, that is totally wrong. You remember Al Gore, right? Remember how silly that guy is? Plus, he lost the election! To George Bush! Lots of smart people don’t believe in climate change and how can the climate change if the earth is flat? Remember how they used to call it global warming? Now these scientists say that some places will get cooler! Also, remember that it snowed in Texas. So much for global warming! And we still had winter; it was cold some days. And everyone knows that we had ice ages in the past. But we don’t have an ice age now. So, climate changes without us; so much for the idea that humans are causing it.”

Moderator: “Time. Your turn Dr. Jones. You have two minutes.”

Dr. Jones: “So where to begin…”

Gus, two minutes later: “See, “Dr.” Jones did not refute all my arguments. So, climate change is all a hoax, as I said.

 

While industrial robots have been in service for a while, household robots have largely been limited to floor cleaning machines like the Roomba. But Physical Intelligence has built a robot that seems capable of doing some household tasks such as folding clothes. While a viable commercial products lie in the future, the dangers of household robots should be considered now. I will skip over the usual fear of the robot rebellion in which the machines turn against humans and focus on more likely dangers.

Like a PC or phone, a household robot runs the risk of software errors, glitches and other problems. While having an app crash on your phone or PC can be annoying, this usually does not put you at risk of physical harm. However, a malfunctioning household robot can be a danger. A viable household robot needs to be strong enough to engage in tasks such as cleaning, folding laundry, and moving objects. This entails that the robot will be strong enough to harm humans and pets. If a robot has a software or hardware issue that interferes with its ability to recognize objects and living creatures, it might try to fold a baby’s clothing while the baby is wearing them or mistake a sleeping cat for clothing or trash and put them in the washing machine or garbage can. Even more concerning is a robot designed to prepare food that misidentifies, for example, a human or pet as the meat to be sliced up and cooked for dinner.

Even laying aside such errors, a home can be a complicated place for a robot to operate in, as there will usually be multiple rooms, different types of furniture, different appliances, as well as various people and pets. This means that a household robot could easily become a hazard (or just useless) simply because of an inability to handle such a complicated and changing environment.

To be fair, these challenges can be addressed in various ways. One option is to limit robots to specific tasks and narrow areas of operation. This might require multiple robots in a home, each assigned to a specific area and set of tasks. For example, a knife wielding kitchen robot might have a fixed location in the kitchen and only be able to slice foods placed within a special  box. As another example, a laundry robot might be confined to a laundry room. Another way to reduce risk is through programming and hardware safeguards. For example, pets and humans might wear devices that provide household robots with their exact location so they can avoid them. This way the robot would not need to depend on visually distinguishing, for example, a cat from a sweater. While things could still go wrong (the ID tag might fail or fall off your cat’s collar), people are generally willing to accept some risk of injury and death for convenience. After all, any electrical appliance in your home can probably kill you and driving anywhere comes with the risk of injury or death. In addition to concerns about accidental injuries, there is also the threat of intentionally caused harm.

Household robots will almost certainly have online connections. On the one hand, this has many potential benefits such as being able to check in on your robots and taking manual control if, for example, one gets stuck in a corner. On the other hand, if you can access your robots online, that means that bad actors can do so as well, just as can happen today with any connected device. The critical difference is that a connected robot in your house means that a bad actor can gain a virtual physical presence in your home and use your robot in various ways.

It is certain that some people will take control of other peoples’ robots just for fun, to do various pranks such as having a robot move things around or make a small mess. But compromised robots could be used for a range of misdeeds, such as unlocking doors (although connected smart locks are obviously vulnerable), grabbing valuables and tossing them out windows, breaking things, and even attacking people and pets. This threat can be mitigated by good security practices but the only two ways to avoid a compromised robot is to either not have it connected or not have one at all.

As with autonomous vehicles, household robots also raise legal concerns about liability. If, for example, your robot injures a guest, there is the question of who has legal responsibility. On the plus side, household robots will be good for some lawyers as this will create a new, profitable subfield of law.

In closing, while the idea of having household robots seems appealing, their presence would create a new set of dangers, especially if they are connected and can be compromised.

 

In utopian science fiction, machines free humans so they can enjoy a life of leisure and enlightenment. In dystopian stories, machines enslave or exterminate humans. Reality has been, on average, a middletopia: a mean between the worst possible world and the best possible world. But a good case can be made that reality is more of a dystopia-lite; a bad world, but better than a full dystopia. While people still dream of utopia, there are those who are working hard to push us further into dystopia.

On a positive note, robots have replaced humans in some jobs that are dirty, dull, or dangerous. In some cases, the displaced humans have moved on to better jobs. In other cases, they have moved into other dirty, dull or dangerous jobs to wait for the machines to replace those jobs.  Machines have also replaced humans in jobs humans see as desirable and AI companies are determined to continue that trend, having selected writing and art as prime targets This leads to questions about what jobs will be left to humans and which will be taken over by the machines

There was once the intuitively appealing view that “creative” jobs would be safe from machines, but physical labor would be easily taken over by machines. On this view, machines will replace jobs such as those held by warehouse pickers, construction workers and janitors. Artists, philosophers, and teachers were supposed to be safe from the machine revolution. In some cases, the intuitive view was correct. Machines are routinely used for physical labor such as constructing cars and robot Socrates has yet to show up. However, the intuitive view about creative tasks is under attack as AI is used in journalism, law, academics and image creation. There are also tasks that would seem easy to automate, such as cleaning toilets or doing construction, that are very hard for robots, but easy for humans.

An example of a task that would seem ideal for automation is warehouse picking, especially of the sort done by Amazon. Amazon and other companies have automated some of the process, making use of robots in various tasks. But humans are still a critical part of the picking process. Since humans tend to have poor memories and get bored with picking, human pickers are “remote controlled” by computers that tell them what to do, then they tell the computers what they have done. For example, a human might be directed to pick five boxes of acne medicine, then five more boxes of acne medicine, then a copy of Fifty Shades of Gray and finally an Android phone. Humans are very good at picking and dealing with things like a broken bottle of shampoo in a box that robots still handle poorly.

In this sort of warehouse, the humans are being controlled by the machines. The machines take care of the higher-level activities of organizing orders and managing, while the human brain handles the task of selecting the right items and dealing with some tasks the machines cannot handle. While selecting seems simple, this is because it is simple for humans but not for existing robots. We are good at recognizing, grouping and distinguishing things and have the manual dexterity to perform the picking tasks, thanks to our opposable thumbs. Unfortunately for the human worker, these picking tasks are probably not very rewarding, creative or interesting and this is exactly the sort of drudge job that robots are supposed to free us from.

While computer-controlled warehouse work is one example of humans being directed by machines, it is easy to imagine this approach applied to tasks that require manual dexterity and what might be called “animal skills” such as object recognition. It is also easy to imagine this approach extended far beyond these jobs as a cost-cutting measure.

One way this approach could cut costs would be by allowing employers to buy “skilled” AI systems and use them to direct unskilled human labor. For simple jobs, a human might be directed via a headset linked to the AI that tells the human what to do, providing the “intelligence” guiding the body. For more complex jobs, a human might wear a VR style helmet with a machine directing the human via augmented reality. For example, an unskilled human could be walked through electrical or plumbing work by an AI. It should be noted that this technology could also be useful for people doing DIY projects and someday a person might be able to rent skills (via AI) as they now rent tools. But this could also impact the labor market, especially if almost anyone could use the technology effectively.

In this system, humans would provide the manual dexterity and all those highly evolved physical capacities. The AI would provide the direction, skill and “intelligence.” Since any adequately functional human body would suffice to serve the controlling AI, the value of such human labor would be low, and wages would match this value. Workers would be easy to replace because if a worker is fired or quits, then a new worker can simply don the interface device and get about the task with little training. This would also save in education costs as AI directed laborer would not need much education in job skills as these are by the AI. Humans would just need the basis skills allowing them to be directed properly by AI. This does point towards a dystopia in which human bodies are driven around through the workday by AI, then released and sent home in driverless cars. One could even imagine this technology being used in education: a human body providing an in-person presence while an AI directs the teaching process.

The employment of humans in these roles would only continue if humans were the cheapest form of available labor. If advances allow robot bodies to do these tasks cheaper, then it would make business sense to replace humans completely.  Alternatively, biological engineering might lead to the production of cost-effective engineered life forms that can replace humans; perhaps a pliable primate that is just smart enough to be directed by the AI. But not human enough to be considered a slave. Or, to go deeper into dystopia, perhaps a cyborg will be built that has hardware in place of the higher parts of the brain and thus serves as a meat robot driven around the job by the AI that is using the evolved biological features that cannot be replicated cost-effectively by machinery. While such things remain science fiction, now is the time to start considering the laws and policies that should govern remote controlled humans in the workplace.

 

Students and employers often complain that college does not prepare them for the real world of filling jobs and this complaint has some merit. But what is the real world of jobs like for most workers? Professor David Graeber got considerable media attention when he published his book Bullshit Jobs: A Theory. He claims that millions of people are working jobs they know are meaningless and unnecessary. Researcher Simon Walo decided to test Graeber’s theory and found that his investigation supported Graeber’s view. While Graeber’s view can be debated, it is reasonable to believe that some jobs are BS all the time and all jobs are BS some of the time. Thus, if educators are to prepare students for working in the real world, they must prepare them for the BS of the workplace. AI can prove useful here.

In an optimistic sci-fi view of the future, AI exists to relieve humans of the dreadful four Ds of bad jobs: the Dangerous, the Degrading, the Dirty, and the Dull. In a bright future, general AI would assist, but not replace, humans in creative and scientific endeavors. In dystopian sci-fi views of the AI future, AI enslaves or exterminates humanity. In dystopia lite, a few humans use AI to make life worse for many humans, such as by replacing humans with AI in good and rewarding jobs.  Much of the effort in AI development seems aimed at making this a reality.

As an example, it is feared that AI will put writers and artists out of work, so when the Hollywood writers went on strike, they wanted protection from being replaced by AI. They succeeded in this goal, but there remains a reasonable question about how great the threat of AI is in terms of its being able to replace humans in jobs humans want to do. Fortunately for humans doing creative and meaningful work, AI is not very good at these tasks. As Arvind Narayanan and Sayash Kapoor have argued, AI of this sort seems to be most useful at doing useless things. But this can be useful for workers and educators should train students to use AI to do these useless things. This might seem a bit crazy but makes perfect sense in our economic reality.

Some jobs are useless, and all jobs have useless tasks. Although his view can be challenged, Graeber came up with three categories of useless jobs. His “flunkies” category consists of people paid to make the rich and important look more rich and more important.  This can be expanded to include all decorative minions. “Goons” are people filling positions existing only because a competitor company created similar jobs. Finally, there are the  “box tickers”, which can be refined to cover jobs workers see as useless but also produce work whose absence would have no meaningful effect on the world.

It must be noted that what is perceived as useless is a matter of values and will vary between persons and in different contexts. To use a silly example, imagine the Florida state legislature mandated that all state universities send in a monthly report in the form of a haiku. Each month, someone will need to create and email the haiku. This task seems useless. But imagine that if a school fails to comply, they lose $1 million in funding. This makes the task useful for the school as a means of protecting their funding. Fortunately, AI can easily complete this useless useful task.

As a serious example, suppose a worker must write reports for management based on bullet points given in presentations. Management, of course, never reads the reports and they are thus useless but required by company policy. While a seemingly rational solution is to eliminate the reports, that is not how bureaucracies usually operate in the “real world.” Fortunately, AI can make the worker’s task easier: they can use AI to transform the bullet points into a report and use the saved time for more meaningful tasks (or viewing social media). Management can also use AI to summarize the report into bullet points. While it would seem more rational to eliminate the reports, this is not how the real world usually works. But what should educators do with AI in their classrooms in the context of useless tasks and jobs?

While this will need to vary from class to class, relevant educators should consider a general overview of jobs and task categories in terms of usefulness and the ability of AI to do these jobs and tasks.  Faculty could then identify the likely useless jobs and useless tasks their students will probably do in the real world. They can then consider how these tasks can be done using AI. This will allow them to create lessons and assignments to give students the skills to use AI to complete useless tasks quickly and with minimal effort. This can allow workers to spend more time on useful work, assuming their jobs have any such tasks.

In closing, my focus has been on using AI for useless tasks. Teaching students to use AI for useful tasks is another subject entirely and while not covered here is certainly worthy of consideration. And here is an AI generated haiku:

 

Eighty percent rise

FAMU students excel

In their learning’s ligh

 

One of the many fears about AI is that it will be weaponized by political candidates. In a proactive move, some states have already created laws regulating its use. Michigan has a law aimed at the deceptive use of AI that requires a disclaimer when a political ad is “manipulated by technical means and depicts speech or conduct that did not occur.”  My adopted state of Florida has a similar law that political ads using generative AI requires a disclaimer. While the effect of disclaimers on elections remains to be seen, a study by New York University’s Center on Technology Policy found that research subjects saw candidates who used such disclaimers as “less trustworthy and less appealing.”

The subjects watched fictional political ads, some of which had AI disclaimers, and then rated the fictional candidates on trustworthiness, truthfulness and how likely they were to vote for them. The study showed that the disclaimers had a small but statistically significant negative impact on the perception of these fictional candidates. This occurred whether the AI use was deceptive or more harmless. The study subjects also expressed a preference for using disclaimers anytime AI was used in an ad, even when the use was harmless, and this held across party lines. As attack ads are a common strategy, it is interesting that the study found that such ads with an AI disclaimer backfired, and the study subjects evaluated the target as more trustworthy and appealing than the attacker.

If the study results hold for real ads, these findings might serve to deter the use of AI in political ads, especially attack ads. But it is worth noting that the study did not involve ads featuring actual candidates. Out in the wild, voters tend to be tolerant of lies or even like them when the lies support their political beliefs. If the disclaimer is seen as stating or implying that the ad contains untruths, it is likely that the negative impact of the disclaimer would be less or even nonexistent for certain candidates or messages. This is something that will need to be assessed in the wild.

The findings also suggest a diabolical strategy in which an attack ad with the AI disclaimer is created to target the candidate the creators support. These supporters would need to take care to conceal their connection to the candidate, but this is easy in the current dark money reality of American politics. They would, of course, need to calculate the risk that the ad might work better as an attack ad than a backfire ad. Speaking of diabolical, it might be wondered why there are disclaimer laws rather than bans.

The Florida law requires a disclaimer when AI is used to “depict a real person performing an action that did not actually occur, and was created with the intent to injure a candidate or to deceive regarding a ballot issue.” A possible example of such use seems to occur in an ad by DeSantis’s campaign falsely depicting Trump embracing Fauci in 2023.   It is noteworthy that the wording of the law entails that the intentional use of AI to harm and deceive in political advertising is allowed but merely requires a disclaimer. That is, an ad is allowed to lie but with a disclaimer. This might strike many as odd, but follows established law.

As the former head of the FCC under Obama Tom Wheeler notes, lies are allowed in political ads on federally regulated broadcast channels. As would be suspected, the arguments used to defend allowing lies in political ads are based on the First Amendment. This “right to lie” provides some explanation as to why these laws do not ban the use of AI. It might be wondered why there is not a more general law requiring a disclaimer for all intentional deceptions in political ads. A practical reason is that it is currently much easier to prove the use of AI than it is to prove intentional deception in general. That said, the Florida law specifies intent and the use of AI to depict something that did not occur and proving both does present a challenge, especially since people can legally lie in their ads and insist the depiction is of something real.

 Cable TV channels, such as CNN, can reject ads. In some cases, stations can reject ads from non-candidate outside groups, such as super PACs. Social media companies, such as X and Facebook, have considerable freedom in what they can reject. Those defending this right of rejection point out the oft forgotten fact that the First Amendment legal right applies to the actions of the government and not private businesses, such as CNN and Facebook. Broadcast TV, as noted above, is an exception to this. The companies that run political ads will need to develop their own AI policies while also following the relevant laws.

While some might think that a complete ban on AI would be best, the AI hype has made this a bad idea. This is because companies have rushed to include AI in as many products as possible and to rebrand existing technologies as AI. For example, the text of an ad might be written in Microsoft Word with Grammarly installed and Grammarly is pitching itself as providing AI writing assistance. Programs like Adobe Illustrator and Photoshop also have AI features that have innocuous uses, such as automating the process of improving the quality of a real image or creating a background pattern that might be used in a print ad.  It would obviously be absurd to require a disclaimer for such uses of AI.

For more on cyber policy issues: Hewlett Foundation Cyber Policy Institute (famu.edu)

 

When ChatGPT and its competitors became available to students, some warned of an AI apocalypse in education.  This fear mirrored the broader worries about the over-hyped dangers of AI. This is not to deny that AI presents challenges and danger, but we need to have a realistic view of the threats and promises so that rational policies and practices can be implemented.

As a professor and the chair of the General Education Assessment Committee at Florida A&M University I assess the work of my students, and I am involved with the broader task of assessing general education. In both cases a key challenge is determining how much of the work turned in by students is their work. After all, we want to know how our students are performing and not how AI or some unknown writer is performing.

While students have been cheating since the advent of education, it was feared AI would cause a cheating tsunami. This worry seemed sensible since AI makes cheating easy, free and harder to detect.  Large language models allow “plagiarism on demand” by generating new text each time. With the development of software such as Turnitin, detecting traditional plagiarism became automated and fast. These tools also identify the sources used in plagiarism, providing professors with reliable evidence. But large language models defeat this method of detection, since they generate original text. Ironically, some faculty now see a 0% plagiarism score on Turnitin as a possible red flag. But has an AI cheating tsunami washed over education?

Determining how many students are cheating is like determining how many people are committing crime: one only knows how many people have been caught and not how many people are doing it. Because of this, caution must be exercised when drawing a conclusion about the extent of cheating otherwise one runs the risk of falling victim to the fallacy of overconfident inference from unknown statistics.

In the case of AI cheating in education, one source of data is Turnitin’s AI detection software. Over the course of a year, the service checked 200 million assignments and flagged AI use in 1 in 10 assignments while 3 in 100 were flagged as mostly AI. These results have remained stable, suggesting that AI cheating is neither a tsunami nor increasing. But this assumes that the AI detection software is accurate.

Turnitin claims it has a false positive rate of 1%. In addition to Turnitin, there are other AI detection services that have been evaluated, with the worst having an accuracy of 38% and the best claimed to have a 90% accuracy. But there are two major problems with the accuracy of existing plagiarism detection software.

The first, as the title of a recent paper notes, “GPT detectors are biased against non-native English writers.” As the authors noted, while AI detectors are nearly perfectly accurate in evaluating essays by U.S. born eighth-graders, they misclassified 61.22% of TOEFL essays written by non-native English students. All seven of the tested detectors incorrectly flagged 18 of the 91 TOEFL essays and 89 of 91 of the essays (97%) were flagged by at least one detector.

The second is that AI detectors can be fooled. The current detectors usually work by evaluating perplexity as a metric. Perplexity, which is a measure of such factors as lexical diversity and grammatical complexity, can be created in AI text by using simple prompt engineering. For example, a student could prompt ChatGPT to rewrite the text using more literary language. There is also a concern that the algorithms used in proprietary detection software will be kept secret, so it will be difficult to determine what biases and defects they might have.

Because of these problems, educators should be cautious when using such software to evaluate student work. This is especially true in cases in which a student is assigned a failing grade or even accused of academic misconduct because they are suspected of using AI. In the case of traditional cheating, a professor could have clear evidence in the form of copied text. In the case of AI detection, the professor only has the evaluation of software whose inner workings are most likely not available for examination and whose true accuracy remains unknown. Because of this, educational institutes need to develop rational guidelines for best practices when using AI detection software. But the question remains as to how likely it is that students will engage in cheating now that ChatGPT and its ilk are readily available.

Stanford scholars Victor Lee and Denise Pope have been studying cheating, and past surveys over 15 years showed that 60-70% of students admitted to cheating. In 2023 the percentage stayed about the same or decreased slightly, even when students were asked about using AI. While there is the concern that cheaters would lie about cheating, Pope and Lee use anonymous surveys and take care in designing the survey questions. While cheating remains a problem, AI has not increased it, and the feared tsunami seems to have died far offshore.

This does make sense in that cheating has always been relatively easy, and the decision to cheat is more a matter of moral and practical judgment rather than based on the available technology. While technology can provide new means of cheating, a student must still be willing to cheat, and that percentage seems to be relatively stable in the face of changing technology.  That said, large language models are a new technology and their long-term impact in cheating is something that needs to be determined. But, so far, the doomsayers predictions have not come true. Fairness requires acknowledging that this might be because educators took effective action to prevent this; it would be poor reasoning to fall for the prediction fallacy.

As a final point of discussion, it is worth considering that  perhaps AI has not resulted in a surge in cheating because it is not a great tool for this. As Arvind Narayanan and Sayash Kapoor have argued, AI seems to be most useful at doing useless things. To be fair, assignments in higher education can be useless things of the type AI is good at doing. But if AI is being used to complete useless assignments, then this is a problem with the assignments (and the professors) and not AI.

In closing, while there is also the concern that AI will get better at cheating or that as students grow up with AI, they will be more inclined to use it to cheat. And, of course, it is worth considering whether such use should be considered cheating or if it is time to retire some types of assignments and change our approach to education as, for example, we did when calculators were accepted.

 

For more on cyber policy issues: Hewlett Foundation Cyber Policy Institute (famu.edu)

While the anti-abortion movement claimed a great victory when the Supreme Court overturned Roe v. Wade, the Republican Party has learned that this victory proved deeply unpopular with the American people. While Democrats favor abortion rights more than Republicans, 64% of surveyed voters say abortion should be always or mostly legal. While some Republican controlled state legislatures have imposed extreme restrictions on abortion rights, abortion rights supporters have won in several state ballots. As this is being written several more states (including my adopted state of Florida) have abortion rights measures on the ballot. Given that the anti-abortion view is held by a minority of voters, it is likely that these measures will pass in many states.

Because the anti-abortion position of the Republican party has proven unpopular and has imposed a political cost, the party’s rhetoric has shifted. The current rhetorical spin is that the Republican party is not against abortion rights. Rather, the party is for states’ rights.  Those critical of this rhetoric like to point out that appeals to states’ rights was also a tactic employed by the southern states to defend slavery. While the analogy is imperfect, the comparison does have some merit.

The states’ rights argument for slavery amounts to contending that the states should have the freedom to decide whether they will allow slavery, and this is usually phrased in terms of an appeal to democracy. That is, the citizens of the state should vote to decide whether some people can be denied freedom and be owned. An obvious defect with this reasoning is that it rests on the assumption that it is a matter of freedom of choice to take away freedom of choice.

A similar defect arises with the states’ rights rhetoric in the abortion debate. If it is accepted that the citizens of the state have the right to decide the issue of abortion because they should be free of federal law, then it is problematic to argue that the state has the right to take away the freedom of women to decide whether they get an abortion. If choice is important, then having legal abortion allows women to choose: a woman is not mandated to have an abortion nor forbidden, so she can make the choice. Hence, this rhetorical move entails that abortion should be legal nationwide.

Someone might counter this by taking the anti-abortion stance that women should not be allowed that choice, perhaps by drawing an analogy with murder. After all, they might argue, we would not want people free to chose murder. But the problem with that reply is that by using the states’ rights rhetoric, the Republican party has acknowledged that the legality of abortion should be a matter of choice, and this makes it difficult to argue that abortion should not be a choice for individual women.

While intended to address the backlash from the unpopularity of the success of the anti-abortion movement, this rhetoric has caused backlash from that movement. Some anti-abortion activists have urged their followers to withdraw their support of Trump. There is the question of how much impact this will have on the election, given that anti-abortion voters will almost certainly not vote for Harris. But it might cause a few single-issue voters to stay home on election day or not vote for Trump.

Pro-abortion rights people are almost certainly not going to be fooled by this rhetoric, since they know this is a rhetorical shift and not a change in policy or goals. While it might win over a few of the undecided voters, it seems to have two effects. The first is that it gives Republicans an established rhetorical talking point to use whenever they are asked about abortion. The second is that it provides those who want to vote for anti-abortion Republicans but who are not anti-abortion themselves a way to rationalize their vote. They can insist the Republican party is “pro-choice” because their new rhetorical position is that the states should chose. But not that women should chose.

The states’ rights rhetorical move could be an effective strategy. While the anti-abortion movement would prefer a federal abortion ban, having the states decide is better for them than having abortion legal nationwide. After all, some states have put abortion bans in place and these have been wins for the movement. But the obvious downside for this movement is that some states have put in place protections for abortion rights, despite the anti-abortion movement’s desire to make the choice for everyone.

In closing, the states’ rights argument is a position that cannot be effectively defended, because its foundation is the principle of choice, and this entails that it is the women who should make the choice for themselves.

 

Description:

This fallacy occurs when someone uncritically rejects a prediction or the effectiveness of the responses to it when the predicted outcome does not occur:

Premise 1: Prediction P predicted outcome X if response R is not taken.

Premise 2: Response R was taken (based on prediction P).

Premise 3: X did not happen, so Prediction P was wrong.

Conclusion: Response R should not have been taken (or there is no longer a need to take Response R).

 

The error occurs because of a failure to consider the obvious: if there is an effective response to a predicted outcome, then the prediction will appear to be “wrong” because the predicted outcome will not occur.

While a prediction that turns out to be “wrong” is technically wrong, the error here is to uncritically conclude that this proves the response was not needed (or there is no longer any need to keep responding). The initial prediction assumes there will not be a response and is usually made to argue for responding. If the response is effective, then the predicted outcome will not occur, which is the point of responding. To reason that the “failure” of the prediction shows that the response was mistaken or no longer needed is thus a mistake in reasoning.

To use a silly analogy, imagine that we are in a car and driving towards a cliff. You make the prediction that if we keep going, we will go off the cliff and die. So, I turn the wheel and avoid the cliff. If backseat Billy gets angry and says that there was no reason to turn the wheel or that I should turn it back because we did not die in a fiery explosion, Billy is falling for this fallacy. After all, if we did not turn, then we would have probably died. And if we turn back too soon, then we will probably die. The point of turning is so that the predicted outcome of death will not occur.

A variation on this fallacy involves inferring the prediction was bad because it turned out to be “wrong”:

Premise 1: Prediction P predicted outcome X if response R is not taken.

Premise 2: Response R was taken based on prediction P.

Premise 3: X did not happen.

Conclusion: Prediction P was wrong about X occurring if response R was not taken.

 

While the prediction would be “wrong” in that the predicted outcome did not occur, this does not disprove the prediction that X would occur without the response. Going back to the car example, the prediction that we would die if we drove of the cliff if we do not turn is not disproven if we turn and then do not die. In fact, that is the result we want.

Since it lacks logical force, this fallacy gains its power from psychological force. Sorting out why something did not happen can be difficult and it is easier to go along with biases, preconceptions, and ideology than it is to sort out a complicated matter.

This fallacy can be committed in good faith out of ignorance. When committed in bad faith, the person using it is aware of the fallacy. The intent is often to use this fallacy to argue against continuing the response or as a bad faith attack on those who implemented or argued for the response. For example, someone might argue in bad faith that a tax cut was not needed to avoid a recession because the predicted recession did not occur after the tax cut. While the tax cut might have not been a factor, simply asserting that they were not needed because the recession did not occur would commit this fallacy.

 

Defense: To avoid inflicting this fallacy on yourself or falling for it, the main defense is to keep in mind that a prediction based on the assumption that a response will not be taken can turn out to be “wrong” if that response is taken. Also, you should remember that the failure of a predicted event to occur after a response is made to prevent it would count as some evidence that the response was effective rather than as proof it was not needed. But care should be taken to avoid uncritically inferring that the response was needed or effective because the predicted event did not occur.

 

Example #1

Julie: “The doctor said that my blood pressure would keep going up unless I improved my diet and started exercising.”

Kendra: “How is your blood pressure now?”

Julie: “Pretty good. I guess I don’t need to keep eating all those vegetables and I can stop going on those walks.”

Kendra: “Why?”

Julie: “Well, she was wrong. My blood pressure did not go up.”

Example #2

Robert: “While minority voters might have needed some protection long ago, I am confident we can remove all those outdated safeguards.”

Kelly: “Why? Aren’t they still needed? Aren’t they what is keeping some states from returning to the days of Jim Crow?”

Robert: “Certainly not. People predicted that would happen, but it didn’t. So, we obviously no longer need those protections in place.”

Kelly: “But, again, aren’t these protections what is keeping that from happening?”

Robert: “Nonsense. Everything will be fine.”

Example #3

Lulu: “I am so mad. We did all this quarantining, masking, shutting down, social distance and other dumb thing for so long and it is obvious we did not need to.”

Paula: “I didn’t like any of that either, but the health professionals say it saved a lot of lives.”

Lulu: “Yeah, those health professionals said that millions of people would die if we didn’t do all that stupid stuff. But look, we didn’t have millions die. So, all that was just a waste.”

Paula: “Maybe doing all that was why more people didn’t die.”

Lulu: “That is what they want you to think.”

 

Since I often reference various fallacies in blog posts I decided to also post the fallacies. These are from my book 110 Fallacies.

Description:

This fallacy is committed when a person places unwarranted confidence in drawing a conclusion from statistics that are unknown.

 

Premise 1: “Unknown” statistical data D is presented.

Conclusion: Claim C is drawn from D with greater confidence than D warrants.

 

Unknown statistical data is just that, statistical data that is unknown. This data is different from “data” that is simply made up because it has at least some foundation.

One type of unknown statistical data is when educated guesses are made based on limited available data. For example, when experts estimate the number of people who use illegal drugs, they are making an educated guess. As another example, when the number of total deaths in any war is reported, it is (at best) an educated guess because no one knows for sure exactly how many people have been killed.

Another common type of unknown statistical data is when it can only be gathered in ways that are likely to result in incomplete or inaccurate data. For example, statistical data about the number of people who have affairs is likely to be in this category. This is because people generally try to conceal their affairs.

Obviously, unknown statistical data is not good data.  But drawing an inference from unknown data need not always be unreasonable or fallacious. This is because the error in the fallacy is being more confident in the conclusion than the unknown data warrants. If the confidence in the conclusion is proportional to the support provided by the evdience, then no fallacy would be committed.

For example, while the exact number of people killed during the war in Afghanistan will remain unknown, it is reasonable to infer from the known data that many people have died. As another example, while the exact number of people who do not pay their taxes is unknown, it is reasonable to infer that the government is losing some revenue because of this.

The error that makes this a fallacy is to place too much confidence in a conclusion drawn from unknown data. Or to be a bit more technical, to overestimate the strength of the argument based on statistical data that is not adequately known.

This is an error of reasoning because, obviously enough, a conclusion is being drawn that is not adequately justified by the premises. This fallacy can be committed in ignorance or intentionally committed.

Naturally, the way in which the statistical data is gathered also needs to be assessed to determine whether other errors have occurred, but that is another matter.

 

Defense: The main defense against this fallacy is to keep in mind that inferences drawn from unknown statistics need to be proportional to the quality of the evidence. The error, as noted above, is placing too much confidence in unknown statistics.

Sorting out exactly how much confidence can be placed in such statistics can be difficult, but it is wise to be wary of any such reasoning. This is especially true when the unknown statistics are being used by someone who is likely to be biased. That said, to simply reject claims because they are based on unknown statistics would also be an error.

 

Example #1

“Several American Muslims are known to be terrorists or at least terrorist supporters. As such, I estimate that there are hundreds of actual and thousands of potential Muslim-American terrorists. Based on this, I am certain that we are in grave danger from this large number of enemies within our own borders.”

Example #2

“Experts estimate that there are about 11 million illegal immigrants in the United States. While some people are not worried about this, consider the fact that the experts estimate that illegals make up about 5% of the total work force. This explains that percentage of American unemployment since these illegals are certainly stealing 5% of America’s jobs. Probably even more, since these lazy illegals often work multiple jobs.”

Example #3

Sally: “I just read an article about cheating.”

Jane: “How to do it?”

Sally: “No! It was about the number of men who cheat.”

Sasha: “So, what did it say?”

Sally: “Well, the author estimated that 40% of men cheat.”

Kelly: “Hmm, there are five of us here.”

Janet: “You know what that means…”

Sally: “Yes, two of our boyfriends are cheating on us. I always thought Bill and Sam had that look…”

Janet: “Hey! Bill would never cheat on me! I bet it is your man. He is always given me the eye!”

Sally: ‘What! I’ll kill him!”

Janet: “Calm down. I was just kidding. I mean, how can they know that 40% of men cheat? I’m sure none of the boys are cheating on us. Well, except maybe Sally’s man.”

Sally: “Hey!”

Example #4

“We can be sure that most, if not all, rich people cheat on their taxes. After all, the IRS has data showing that some rich people have been caught doing so. Not paying their fair share is exactly what the selfish rich would do.”