Jokes about AI regulation? Ethicists, lawyers and supervisory authorities in particular could get upset about this. Especially if the jokes were created with a GPAI tool. Below are some reasons why humorous content on AI regulation has nevertheless become a (albeit controversial) section of CAIR4.

“Laughter kills fear”

To start with a quote:

“Laughter is a weakness, a decay, a dissolution of fear. And what dissolves fear better than laughter? [ … ] Laughter combines stupidity and joy, two very dangerous things that distract us from true knowledge.”

Old Jorge from “The Name of the Rose”

Humor is an ambivalent topic for many people. In the context of AI regulation and the emerging risks of AI, the meaning of humor can and must also be examined particularly critically. However, one should not overlook the important advantages of humor. Also in the context of AI regulation. More on that later.

1. The CAIR4 section “Humorous”

First of all, the background of this article:

• A new content area has been added to CAIR4: “Humorous”.
• There you will find humorous interpretations of the technical articles published on CAIR4.
• All related text and image content was created using ChatGPT 4o.

The link to the section can be found here and in the main menu under “Contents”:

The jokes about AI regulation presented in the section are all based on the same consideration:

• CAIR4’s technical contributions are serious and usually very specific – not least because AI regulation as such is highly specific. And it’s not really exciting and not funny at all.
• The technical contributions are sometimes so specific that their operational relevance within the highly complex AI world cannot always be conveyed quite so easily.
• According to the view represented here, humor helps to clarify the core message of the individual contributions in an application-oriented way and to arouse curiosity.

And how are the jokes created?

• Operationally, a single CAIR4 technical article is first entered manually into ChatGPT.
• This is followed by the request to create a joke from the post and generate an accompanying image.

All jokes are taken over by ChatGPT except for minimal changes. The AI is therefore the originator of the humorous preparation. Nevertheless, every joke is based on a CAIR4 technical contribution that is as complex as it is serious. In almost all cases, the jokes concern the correct understanding or feasibility of the EU AI Act. In this respect, it is quite interesting per se whether and to what extent generative AI is able not only to understand the content of an article on AI regulation, but also to put it into a humorous context.

Whether and to what extent this will succeed? Everyone can see for themselves. Many of the jokes make you smile – not necessarily laugh. Especially when you keep noticing little breaks in the jokes – ones that make it clear that GPAI is not quite perfect when it comes to humor. And it’s just like in real life: Some jokes only become really funny when the punchline fails – whether due to poor storytelling or simply a misunderstanding. Then the situation is funny, not so much the joke as such.

2. Spot about AI is allowed – and often

The many small and large errors of GPAI are repeatedly mocked (without any reference to regulation). There are countless sites on the Internet where people make fun of cases in which an AI is either “brilliantly wrong” or brings other curiosities to light.

Here is a small selection (German):

• Ridicule for false search results
• Joke collection about Artificial Intelligence
• Nagging about AI-using companies

This kind of spot or joke is usually at the expense of the AI manufacturers, the many “fans” of AI and other proponents. It suggests that people are still the measure of all things. Conversely, the AI would be error-prone, outrageous or, in case of doubt, ridiculous.

Such statements are without question permitted. In the criminal sense, formulations are even possible in the case of abusive criticism in this regard, which could be interpreted as an insult in the case of people (even mere laughter at other people can touch the boundaries of insult in individual cases – at least in germany).

3. Jokes and spot about lawyers

Whether funny or not: Over time, legal jokes have also gained some popularity. Quite a few lawyers tend to be self-ironic by spreading corresponding jokes. Here are some (german) examples:

• Listing on a law firm’s website
• A portal of a habilitated and doctorate lawyer with the URL Juristenwitze (= lawyer jokes)
• The jurawiki portal  with various legal jokes

Many of the jokes have a common core:

• The often unpredictable decision of a judge,
• the ambiguity of laws,
• the risk of over-regulation, or
• the principle of helplessness in the face of formal law.

In this respect, many lawyers in particular take themselves for a ride – they too sometimes suffer from the weaknesses of the legal profession. In this sense, they are “perpetrators” and “qualified victims” at the same time.

4. Jokes about AI regulation issues

This leads to the topic at hand: jokes about AI regulation. This is a special topic in several respects:

1. AI regulation is uncharted territory – in this respect, many regulations are based on facts that are sometimes unclear or difficult to penetrate. Mistakes are inevitable.
2. AI innovation is incredibly fast – in this respect, AI regulation has to work with sometimes vague terms in order to be open to the future.
3. AI seems human-like in many respects – in this respect, it is also able to communicate about AI regulation itself – and generate jokes about it.

With regard to the last point, there is something more important!

• The humorous abilities of AI are even validated in comic competitions.
• According to the study, some generative AI masters humor better than humans.
• Conversely, such a comparison is  considered ridiculous by others.

What is really true or not in this respect may remain unanswered. However, it is obvious that some GPAI models are able to interpret humor and create humorous content based on training data and certain AI techniques. Also in the role of the object affected by AI regulation!

5. Help in difficult situations

However, the task of humor is not always to make fun of others or to make people laugh. Humor also has the important task of making problems and difficult situations “more digestible”:

• During the First World War, for example, soldiers wrote jokes on cave or bunker walls .
• In the German health sector, the most absurd anecdotes were sought.
• Humor is an important tool, especially in times of crisis.

“Black humor” is an effective, but also highly ambivalent topic, especially in difficult situations. There are many critics as well as supporters of it, and rightly so. It depends on the individual case!

6. Overcoming blockages

Another important quality of humor is that it helps to overcome blockages. And this is precisely one of the central challenges in the case of AI regulation: It is a matter of maintaining the willingness to learn in a rapidly changing world.

“Overcoming the fear of change with humour”

AI innovation and with it AI regulation mean almost endless change! Somehow, almost everyone has to change the already gained and sometimes laboriously acquired habits with regard to AI sooner or later. Today. Tomorrow. The day after tomorrow. The deadline model of the EU AI Act still grants some postponement in some respects. But the change through AI regulation is coming – it has already begun! And this change must be made bearable.

And on both sides:

• The regulating side and
• on the part of those affected by regulation.

Especially with regard to the implementation of sometimes quite vague specifications, it is important to maintain the (“gallows”) humor on the part of the AI economy. However, supervisory authorities will also be required to accompany the implementation of the EU AI Act with presumably small and not always fully qualified resources. In view of many “vaguenesses”, the regulatory side can also demonstrate appropriate self-irony with a wink: The EU AI Act is without question an incredibly important step in the right direction. Nevertheless, the law was knitted with a hot needle for many reasons. It goes without saying that this inevitably leads to gaps or inconsistencies. Pointing out the potential for optimization in this regard in a humorous way is therefore not an insult to lèse-majesté – on the contrary: it serves to improve the good.

7. GPAI as a “biased” joke teller?!

The last question that arises in this context: If humor is necessary on both sides, and the AI itself is usually on the side of those affected, how can you leave the field of the joke teller to it?

• Presumably, it would be the case among people that the tendency of those affected by regulation to self-irony or self-criticism is limited.
• On the other hand, there is a tendency to always tell jokes at the expense of the other side: In the case of the joking GPAI, this would be regulators and less the AI economy!

However, this is only a problem to a limited extent: the focus is always on binding the GPAI tools to the specified technical articles. Consequently, the AI does not invent problems that it makes fun of on its own initiative. Therefore, the humorous elements are impartial.

It must also be said quite pragmatically that not every one of the “humorous” stories generated with GPAI is a real “killer joke”. Some of the anecdotes are rather profound. Sometimes they contain a philosophical conclusion – not necessarily a criterion for a good joke: in this respect, the section is also an experiment in the capabilities of generative AI with regard to the regulations of the EU AI Act on art and satire. The images generated for the AI content are also very revealing: is the humor of the text also reflected in them? This is often the case.

See also the following article on the EU AI Act, art, freedom of expression and satire (so far only in German):

8. Conclusion

The integration of humorous, GPAI-generated summaries of the CAIR4 technical articles is quite a balancing act. There are various reasons to discuss this step controversially. In the sense of pro and con. And in order to be able to discuss jokes about AI regulation controversially, you have to publish them first. Exactly!

In this context, the feedback from google-gemini on the topic “jokes on AI regulation” seems revealing. It was worth reading this post as a comment in the original one way or another – in this respect, the AI should also be granted “the last word” or the “first comment” in this matter!

And here’s the link to the GPAI-generated humor on AI regulation again:

About the Autor:

Oliver M. Merx