Page 34 - aips-IV-23.indd

COVER STORY
staff in recent years, and the money seems to have been redirected for an easy solu- tion, an e-commerce company that works with retailers and publishers. Tempta- tions of linking content, audiences, al- gorithms and e-commerce have always led to unethical shortcuts. Why should it be different with AI? Of course, Advon Commerce did not make any statements about the AI scandal.
Dozens of parachuters like them are try- ing to land on fertile soil every day. As an example, there are dozens of spam emails arriving every day to my AIPS address. Most of them offer the trick of some kind of optimisation using AI. Literally speak- ing, scams have also evolved.
A few years ago, companies –including media outlets in difficulty– were seeking some magical recipe to stay afloat. Instead of looking to their own employees and in- vesting in quality reporting, they needed some urgent by-pass. Normally coming from an outsider, these third-parties with magical solutions claimed they can offer the tools and strategies to monetise (that was the keyword to grant their meetings with CEOs and publishers) what they weren't able to achieve with the existing staff. Almost as the email claiming that you’ve won the lottery, but with more fashionable words.
Now, the scam is still there, but the key- word is to have an AI something. As it is some kind of relief. The robot is work- ing. Our algorithm is efficient and prob- lem-solving. It will cost you money, but you will be able to earn more once the optimisation process is over...
Who would be so courageous to invest in some real, old-fashioned way of reporting when you can have a virtual correspond- ent for less money? Tempting, right? Well, just as it happens with human brains, there are also a lot of different AIs operating simultaneously. Therefore, the sole mention of AI does not, in fact, mean anything. It could be excellent. It could be awful.
Think 25 years ago, when Netscape’s Navigator and Internet Explorer were fighting the so-called browser wars, while the search engines were also grasping at the chance of having a bigger piece of the market, with Altavista, Yahoo! and Google quickly emerging as contenders
and enemies. Countless of other browsers and search engines, even from hot shots like Microsoft, perished in the process. It won’t be different in the AI world. Just as it happened with their own websites, any media outlet willing to custom-made their own AI might suffer from the lack of proper updating, and will have to heav- ily work with system engineers on secu- rity and continuous optimisation. Since there are not two different buckets for budget, but only one, where will most of the money willing to be invested go? To journalists or to engineers? Take your guess. To invest in the wrong kind of AI, for example an AI that will become obso- lete compared to others, can have devas- tating results for small media outlets. Are we really sure that what is supposed to be cheap isn’t going to be a financial bomb? The Trojan horse, again.
AI AS A TOOL At the same time, there’s the view from the individual perspective. The freelancers and employees that can, po- tentially, benefit with the good use of AI. Just as it happened with search engines, the fact of knowing how to use it, where and for what, can certainly make a differ- ence. It seems spot-on to say that AI won’t replace you, but only if you know how to use AI you won’t be replaced.
But, as we proved in an eye-opening midnight Zoom call with AIPS president Gianni Merlo, the big risk is to leave the tasks of being a journalist in the hands -or wires- of a robot. While using Chat GPT and another AI website, ProfG, we tried to understand what had happened in the men’s 100m final of the 1972 Olympic Games.
The results, a total dismay in terms of the 5Ws of journalism, were focused on USA’s Eddie Hart and Rey Robinson, rather than on the actual winner, Russian Valery Borzov. Hart and Robinson didn’t compete in the final because they missed their bout. They were late.
It took us six questions, correcting ChatGPT, and each time it delivered a new, obscure version of the events, focus- ing on controversy and using the words “disqualification”, which was not correct. Not even by specifically asking to focus on the winner could we achieve the desired chronicle. Truth is to say that the chatbot
was very polite and apologised, but every time it wrote a perfectly well-elaborated version, it was, indeed, not true. But how many will recall this, and not just blindly trust in what the chatbot is saying? In the Young Reporters Programmes, we have continuously used the same introductory line: Wikipedia is not a valid source. Well, it’s the same with AI, it seems.
This is particularly evident when asking Chat GPT for a profile of any athlete. Bias will be there. Lack of precision. Twisted facts. But all will sound very coherent, even with fabricated information. If you are testing the machine, just like a teacher would test the pupils, then the conversa- tion with the chatbot will become a good article itself.
When asking about Eric Moussambani and his appalling 100m freestyle perfor- mance in Sydney 2000, the first answer led to “one of the most inspiring and un- forgettable moments in the history of the Olympic Games”, with a time of 1m52s, “significantly slower than most profes- sional swimmers”. Even after asking sev- eral times, the chatbot couldn’t say if this was the worst time in the history of that discipline, and after apologising, said it was “one of the slowest”.
NEXT LEVEL OF FAKE What’s really wor- rying, in the era of misinformation, is how AI can be the tool indirectly capable of destroying what was already written by actual journalists, not bloggers, not users, not content-creators, but real journalists. If, using AI, you can write and publish almost anything, the quantity of AI-gen- erated content that will be published can be much larger than the ones written by humans. If other AIs, while searching for answers, find these texts written with mistakes by other robots, decide they are true and use the information to write or rewrite the course of events, that means that we will really be in the feared era of fake news.
Only a few will be able to unplug from the Matrix. The witnesses of history, the jour- nalists, will have to fight against what the robots want to establish as truth, because it’s what the audience wants.
The fake news from the era of made-up screenshots and Photoshop are well over. Now, the fake news has audio messag-
34 | AIPSMAGAZINE