Blog Random by Robert Pepper 18 January 2023
Has ChatGPT made journalists redundant?
We’ve had revolutions before. At some point in the distant past, Grog and friends, living in temperate climates and on a chilly night, left their cave and went to investigate a big red crackling apparition…and got warm. Later, Grog’s descendants accidentally dropped some meat in the fire, noticed it a bit later, decided to taste it…and today we have Maccas.
Then there’s the wheel. Humanity’s greatest invention? No, I’d argue that was writing, as it made inter-generational and inter-community knowledge sharing possible in ways verbal communication couldn’t ever match; writing is more accurate than a verbal retelling, and multiple copies of a written document can be distributed.
At some point our nomadic ancestors noticed that seeds dropped into broken earth grew into plants, and thus agriculture was born, the genesis of what was perhaps greatest change to how humans live, allowing greater population density, and creating time for people to reflect, invent and experiment like never before which led to technological progress as well as war, disease and a reliance on food sources which when failed led to widespread famine. We also figured out how to change crops for our purposes, creating barley and corn for example, and domesticated animals to make them into horses, cows, dogs and cats, all of which are effectively man-made over centuries.
And now here we are in the technological age, with artificial intelligences. Enter ChatGPT, perhaps the most advanced AI made available for public interaction. The possibilities are endless, and like all new tech, there will uses and consequences nobody can foresee.
But I am today interested in a rather more narrow view of how good AI is than the future of humanity – specifically, could it replace me as a specialist automotive writer? The answer is, sadly, not a definitive “no”, and of course that’s today in January 2023, not January 2024 or 2025. I and the other humans won’t get much better, but AI certainly will. A good example of that is chess playing – leaving aside the question of whether chess is AI or not, the point is that it took a while for a computer to beat the best in the world, and now no human can compete. Humans evolve far slower than computers.
Anyway, in the video below I put a few car questions to ChatGPT and analyse the answers. The key thing for me is that every answer looked credible, even if it was wrong and occasionally, self-contradictory. The other key point is that often I felt ChatGPT’s answers missed key information; what was there wasn’t necessarily completely wrong, but there were really good obvious points missed, or the points given were in the wrong order. So, I feel I can do a better job…but that’s not as important as you may think.
If all you want is quick content, you don’t much care whether it’s right or wrong. The industry is already heading that way, as we don’t have technical editors or fact checkers any more, if ever they have existed for automotive at least. There’s a reliance on the journalist to get it right, and the journalist generally has no training nor qualifications. So, we make mistakes too, and more so with today’s pressure to churn out words for clickbait. ChatGPT can produce a lot of content that looks good, for almost no cost, almost instantly. It’s hard, if not impossible to compete with that, as while my writing may be better (for the moment anyway), it’s not as cost-effective. Editors could now say to ChatGPT “write me 5 sand driving tips” instead of paying me to do the same.
The technological revolution with AI is, I believe, the most significant change in human history. It’s both terrifying and exciting to be on its cusp.
Finally, it’s not so much artificial intelligence as the problem, more the impossibility of writing algorithms which don’t have unforeseen negative consequences, or that humans cannot circumvent. From Dilbert: