On Moloch and do people like the songs they have already heard.

Moloch is the Phoenician god of fire, to whom, according to myth, children were sacrificed. Children were placed on Moloch's statue, from where they fell into the abyss of fire.

In the Middle Ages, Moloch was depicted as one of the demons of hell and thus made his way into today's mass culture as a character in B-grade horror films.?

But he has also been lucky enough to find his place in high culture.?

Allen Ginsberg, a poet of the beatnik generation in the 1950s, in his poem 'Howl' made the Molloch into an industrial civilisation that devours everything it encounters in its path and that mankind has so far created.

A fortnight ago, a high-profile Open Letter appeared, signed by well-known names from the technology industry as well as intellectuals (Elon Musk, Steve Wozniak, Yuval Noah Harari) calling for a halt to the training of AI large language models for a period of six months.

I had a rather sceptical attitude to this idea from the start, but I was all the more interested in Lex Fridman's conversation with one of the signatories of the letter Max Tegmark, a prominent physicist and AI scientist whom I hold in high regard. I was hoping to learn in detail about the arguments of the signatories, as I am always very interested in the thinking of those with whom I intuitively disagree. I am also not a person who cannot be persuaded by good arguments to change their position.

But here I have to admit that I was disappointed by Fridman's conversation with Tegmark.

I took careful notes of the discussion for myself, but when I now sat down to write this article, I find it difficult to piece together from these notes any kind of whole that would present a coherent line of reasoning from the opponents of further coaching of LLMs. Let's try, however.

Throughout much of the interview, the axis of Tegmark's argument is the analogy of the race to develop AI by corporations and states to Ginsberg's myth of Moloch.?

Moloch, in Ginsberg's poetry, is a monster that sets people up in a 'game to the bottom'. Everyone incorporating AI into their operations, be it the Big Five of Big Tech or the global powers, are aware of the risks of AI according to Tegmark, but no one will give up those risks knowing that others will not either.

AI in the hands of big corporations (and increasingly big tech) is Moloch. And in the hands of countries such as China, it is Moloch to the powerful.

Given the exponentially increasing development of LLMs, they are becoming a Moloch that is unstoppable.

According to commentators (here I am already going beyond the mentioned interview):

Moloch causes us to automate work without looking at the human cost, because "others do it too”. Moloch will kill the white collar labour market. Here the counter-argument is the masses of analyses of the Industrial Revolution, which show how unemployment rose in the short term and then fell. I recommend books by Thomas Piketty and Richard Susskind. For now, the labour market globally is record strong, although there has been talk of an economic slowdown all the time since the Pandemic.

Moloch means that we can create any deep fakes we want and use them for all sorts of (not necessarily good) purposes. And in the case of the world's superpowers and their intelligence services, it may not just be pictures of the Pope in a Balenciaga puffer. Counter-argument: We underestimate the role of the opinions of traditional, not social, media. I also look forward to the development of cybersecurity, but that's a thing for a separate longer topic.

One line of Tegmark's argument, however, I do agree with and find very relevant and pertinent.

ChatGPT creates in a 'minimum viable way'. His content is therefore devoid of what we value most in man-made texts - personal reflection, the play of associations, the colour of words. ChatGPT would say more dryly, though perhaps more precisely-it kills individualism.

I myself tried to use ChatGPT and stopped for this reason.

And yes, the kind and style of communication that ChatGPT promotes does shallow the exchange of ideas between people.?

But hasn't the advent of social networks and smartphones radically shortened our speech? It certainly has.

And has it, by any chance, made our thinking more shallow already?

Humans are by nature intellectually lazy.

Will Moloch make us stop being creative?

If we look at the intellectual history of humanity (apologies for my Western-centricity) over the last 2000+ years there are intertwined periods of intellectual springs and winters. And what are and were intellectual winters? I would call them periods of intellectual 'copy+paste'.

Starting with Rome, which used the intellectual achievements of Greece like philosophy, poetry and art. According to some, Rome did little by creating itself. But it nevertheless created such an invention as the law of contracts.

But the epoch of Greece and Rome together, as the classical epoch, is contrasted as an intellectual springtime with the winter of the Middle Ages - ?the dark ages”.

The Middle Ages were followed by the intellectual explosion of the Renaissance, while it was later followed by the Baroque era, which is regarded as an example of absolute "copy+paste" or, if you prefer, absolute "mesh up".

But did the eras of intellectual winters really contribute nothing to overall development?

Take the Middle Ages, for example. To this day we use the adjective "medieval" in a pejorative sense to emphasise how backward something is.

Yet, the Middle Ages were the era of great Christian philosophers such as St Augustine and St Thomas Aquinas.

Scholasticism was a method of textual interpretation developed in the Middle Ages, including the interpretation of the philosophy of the ancient Greeks. Later in the Renaissance, the term took on a pejorative meaning as the holding of disputes over contrived problems, but scholastics developed the rational analysis of the text and combined classical and Christian philosophy. Without scholasticism there would have been no Renaissance.

"The 'Dark Age' of the Middle Ages was also the heyday of monasteries and convents, where monks transcribed books. And at the same time they studied them. This can be seen as the prototype of the universities that emerged in the Renaissance era.

So there was also a lot going on in the Middle Ages in terms of intellectual development. It was not just the era of Moloch.

If now we also have a "dark age", because AI, instead of encouraging us to constantly create, as it were, drowns us in a constant, unfettered "mesh up", I intuitively feel that something good will come of it after all.?

I see gigantic potential for LLMs in the development of medicine as we do not know it yet, but that too is a topic for a separate entry. I see no reason to delay it.

I refer you to Reid Hoffman's book 'Impromptu' to see how LLMs can radically aid the learning process and education.

And back to the main thread. A gigantic change is taking place at a level: how one works intellectually.

Until the invention of generative AI, we worked 'directly', so to speak. Just as now, as I write this, we were pouring the product of our imagination into text, drawing or code. Now we do it "indirectly". And the intermediary is our digital agent - generative AI. When I realise this, I do indeed feel uncomfortable. But didn't people feel the same way after the invention of printing? The ubiquity of knowledge that slowly began to happen, which until then had been manually transcribed, must have really hit. A new model of intellectual work would have to be learned. Just like people learnt to use books.

And as for Moloch and myths. My Wife showed me a very interesting story. In the 1980s, Disney was in crisis. None of its productions were catching the market. Until Chris Vogler found himself.?

This young Disney employee came across mythologist Joseph Cambell's work 'The Hero with a Thousand Faces'. He realised that, based on it, George Lucas had created the concept of the ideal myth hero and applied it to Star Wars. With commercial success.?

This is how 'The Lion King' then became the highest grossing film in history. Myths are and have always been with us. And as you can see, they also have enormous commercial power. And we don't need ChatGPT4 or subsequent versions to process them over and over again.

As a character of a cult Polish comedy from the 1960s says: "You know, I like the songs I've already heard".

If you want to make your judgment "on Moloch":