C: Stock-Asso / Shutterstock

Will AI ever write a funny buke?

A long time ago I was correcting essays on the Thirty Years War in St. Patrick’s, Drumcondra – now part of Dublin City University (DCU).  It is a complex subject and when speaking about it I would dread being asked about the finer details of the Edict of Restitution.

These were the earlyish days of the interweb but already water had found its own level and there were people selling essays on many subjects to the more lazy – or perhaps enterprising – students. So, my initial favourable impression of one essay was somewhat marred when I came to the concluding paragraph which began something on the lines of “So, as we look back on the terrible events of the Europe of the middle decades of the 17th century from the perspective of Fordham University in the 1980s ….”

Mmm. Not only had the enterprising student – possibly now teaching in a primary school near you or perhaps a more worldly wise Bitcoin millionaire basking in the shade on a beach in Mauritius – been guilty of plagiarism, but had obviously not bothered reading the thing before presenting it to this eijit.

Anyway, we had a laugh about it when I handed it back, stamped with the equivalent of a C-  and a remark that the presenter would have been well employed in the courts of the devious Princes of the Heilbronn League rather than aspiring to be a Múinteoir.

I was reminded of this by the ongoing controversy around Artificial Intelligence, or AI. Intleacht Shaorga as Gaeilge, from saorga – artificial, which interestingly according to Gary Bannister’s magisterial Teasáras Gaeilge-Béarla has synonyms which include áiféiseach – “contrived; ridiculous”, bréagach – “false”, cumtha – “concocted,” and even móiréiseach – “pretentious.”

Which perhaps sums up the ambivalence surrounding AI that is common to most cultures although I have seen one English language thesaurus which appears to take a more benign view of the full term which is given synonyms such as “natural language processing,” and “knowledge engineering,” although the latter can be regarded as either good or bad, depending on one’s perspective on such matters.

“Knowledge engineering” would probably be an apt term for the historical and scientific and other intellectual distortions that are common to totalitarianism which seeks to engineer knowledge and reality to fit the purposes of the state by basically persuading people that what they think they know is false, and what they think they know is false, is in fact the truth.

It may be no coincidence then that the Chinese Communist state is to the forefront of the development of Artificial Intelligence. One of the aims of the current five year plan is for China to become the world leader in AI by 2030, by which stage that would be worth somewhere in the region of €200 billion a year, equivalent to an annual increase of over 1% in the economy as a whole.

This has led to concerns over the use of AI in systems such as facial recognition which are a key part of the claustrophobic control that the Chinese state exercises over its own citizens, and the benefits of which they are attempting to sell to clients in the rest of the world.

The Chinese state authority has declared that AI must “reflect the core values of socialism,” and given what we know of these from its murderous history, we may pause to reflect on its benefits. Especially given that Chinese companies such as Baidu, Huawei and Alibaba are to the forefront of AI globally.

People here have become more aware of the potentially malign, if amusing, impacts of Artificial Intelligence on the public discourse by the recent publication by the Irish Times of an AI generated piece on the “problematic”, as in racist, implications of Irish women using fake tan.

Arguably the most absurd aspect of all of this was not that the news outlet was gulled by a machine, but that its editorial team thought that this sort of tendentious nonsense was worthy of publication no matter who was responsible for it.

Which begs the question as to how difficult might it become to distinguish between things generated by AI and things genuinely the creation of human beings.  Which is complicated of course by the fact that AI itself is the creation of human beings and that it is programmed by software writers and engineers.  That it might ever become independent of human agency is therefore still in the realms of science fiction along with robots and matrixes which one day decide of their own volition to take over the world.

Actually the fact that AI is the creation and under the control of human beings is far more worrying than that it might become autonomous. Especially if the controllers of AI are giant corporations and states who may not have the best interests of the rest of us at the heart of their project.

The focus of public debate, however, remains how we might be able to distinguish between Artificial and Human Intelligence.  To go back to my opening anecdote about the essay downloaded from the interweb, how might it be possible to decide whether an essay or a song or a photograph or whatever was the creation of a machine?

Someone made the point recently that this will always be possible as the “machine” was ultimately dependent on the input of the human seeking perhaps to con someone with an essay for example, or dependent in the case of music or creative writing or visual art on previous creations which it possesses the programmed ability to replicate and synthesize.

But, that still leaves me thinking how you would be able to tell the difference between a Bob Dylan song from his Blood on the Tracks period, and a song that a computer was asked to write that sounded like a Bob Dylan song from 1974, and even one that sounded like it was sung by Bob Dylan and recorded on vinyl by Mr. Zimmerman in 1974?

Which led me to think, well surely AI cannot replicate humour, or even nuance?  Could a computer be successfully programmed to write something like Flann O’Brien’s An Béal Bocht or John Kennedy Toole’s A Confederacy of Dunces that would be equally as surreally funny? And of course in common with most things I think of, someone else has thought of it first.

Researchers at Standford University – in cooperation with Google which cannot be a good thing – have been working on “teaching” computers to “think” more like humans by feeding them with thousands of novels. This has led to the production of some bad poetry.

A Chinese company has even managed to have AI produced poetry published and led readers to believe that it had been written by a person.  (No smart remarks now about the Irish Times cultural gurus.)

The one I have seen is dreadful – although that is never a bar to publication and even fame – and crucially it is not remotely funny or suggestive of any human quality really.  Which some analysts have attributed to a computer’s failure to grasp the nuances of human psychology, never mind humour.


 “The rain is blowing through the sea

A bird in the sky

A night of light and calm


Now in the sky

Cool heart

The savage north wind

When I found a new world…”

So there you go. All I need now is to find a computer that can do Matt Treacy pastiches and I can go back to shopping for early afternoon pints of porter.

Share mdi-share-variant mdi-twitter mdi-facebook mdi-whatsapp mdi-telegram mdi-linkedin mdi-email mdi-printer mdi-chevron-left Prev Next mdi-chevron-right Related
Comments are closed

Do you support the Governments plans to put calorie labels on wine bottles?

View Results

Loading ... Loading ...