In the long term I think AI has the potential to revolutionize journalism. To get there it has to make a series of advances that render the criticisms in thi

In the long term I think AI has the potential to revolutionize journalism. To get there it has to make a series of advances that render the criticisms in this article moot. It cannot make things up, and it cannot be wrong, ever. There is distrust in the media, much of that is for good reason. There have been countless examples of news stories that lie, misrepresent, obfuscate or simply neglect important information. That those stories exist places all stories in some degree of doubt. The lack of trust comes not from disinformation, that's always been there, but the awareness of disinformation. The only thing that people will believe is a news source backed by a reputation, Reputations take ages to build and seconds to shatter. There is a huge scope for machine learning to safeguard against the incidents that shatter a news organizations reputation. Humans are too fallible. There is a chance that AI might not be. The author bases his argument on the presumption listed under the heading Why generative AI is not the answer What follows is the usual tired list of criticisms, starting with the ever so tired 'designed to predict the next word in a sentence'. When people hear this what they imagine is something like a Marcov-chain. That is not what LLMs do. If you want to argue that predicting the next word in a sentence (given massive context) cannot lead to whatever you say the AI will not be able to do, devise a task that cannot be completed by performing next word prediction. Bear in mind that LLMs also add to their own context as they generate output creating complex patterns of behavior. If you wish to deny that architectures made of many smaller, simpler, components can be more than those simple features, then you are going to have to take a serious look at brain cells. Yes people are concerned, and yes AI's are not that great, their interfaces are not very good right now. The same could be said for Cars or any number of technologies as they emerged. There are some optimists who see an AI analogous to Pong(which is where we are right now) and can imagine an AI analogous to Crysis. They might be over-optimistic, but denying that there is any possibility of it seems downright foolhardy.