Wikipedia worries

Yesterday afternoon, I was diverted onto A level examinations, turning up reference 1 and resulting in the postscripts to the post at reference 2.

Later on in the day, I thought I would ask Bard what he know about the history of the A level examination and he was able to spout convincingly, particularly when prompted, although I have not yet got around to checking any of his details and it may well be that he has made some of them up. And I also worry that Bard puts too much weight on prompts and tailors his answers around them. He says what he thinks you want to hear. But all that apart, his efforts were far inferior to that at reference 1 - to which he said, when asked, that he did not have access.

Furthermore, reference 1 gets a level of trust which I do not accord Bard. First, the TES is a respectable, long established weekly magazine. Although that said, while at the time of writing I think it was owned by the not-so-respectable NewsCorp, it has now been sold to Providence Equity of reference 3. Second, the article was written in 2014, hopefully before computers starting muscling in on journalism. Third, although I have not been able to dig out any biographical details for this William Stewart, he does write a lot of pieces for the TES and might be presumed to know about education. Probably not the one who appears to come from the US and is now a professor of education in Korea.

The trouble is that while I do not trust Bard - or his friends and relatives - he can write decent, plausible prose and he will no doubt improve over time. There seem to be an awful lot of people out there beavering away to make this happen. An ocean of learned papers about large language models. An army of highly qualified computer scientists busily chopping off the branch on which they are sitting. Funny things, people!

Which brings me onto Wikipedia, largely staffed up by another army, this one of unpaid editors, an army made up, I dare say, of people from all kinds of backgrounds.

But what is going to stop some of these editors turning to the likes of Bard to generate their copy - complete with a raft of plausible but invented detail? Perhaps turning to the likes of Bard without even knowing it, with Microsoft's Bing now seeming to incorporate something from the Bard stable. And perhaps some of these editors, on the assumption that no-one will check, will go so far as to fake up some plausible looking references for that detail. 

In short, as things stand, I trust Wikipedia and I have only rarely come across an error therein. To me it is a very valued resource - and to which I might say that I contribute a modest amount of money, if not time.

The worry is that going forward, Wikipedia management and the army of unpaid editors will not be able to keep the likes of Bard at bay. That one will no longer be so confident that Wikipedia, if not the whole truth, is nothing but the truth. Or at least near enough. There is bound to be some error in an enterprise on this scale. No system is going to be perfect. But is it about to become a lot less perfect?

PS: similar considerations apply to a large proportion of the stuff which finds its way onto the Internet. But Wikipedia is often my first port of call when broaching a new topic.

References

Reference 1: Feature: the long history of the A-level – William Stewart, TES magazine – 2014.

Reference 2: https://psmv5.blogspot.com/2023/10/trios.html.

Reference 3: https://en.wikipedia.org/wiki/Providence_Equity.

Reference 4: https://en.wikipedia.org/wiki/Wikimedia_Foundation. The fount of all wisdom.