Will A.I. Kill History?

The answer is definitely yes if the doomsayer futurists are to be believed. They are predicting that A.I. (Artificial Intelligence) will lead to ‘a 99.9% chance of human extinction due to A.I. within the next century.’

The answer, however, is probably or possibly, if the more moderate futurists are to be believed.

History has been something that has never been static. The study of history, the sharing of history (for example, the teaching of history) has evolved over time…and it is still changing.

Not too many years ago, if you were to research and then write an article about a history topic – let’s say, about Aboriginal breastplates – you would probably have spent many, many hours trawling through paper-based material in numerous libraries or museums.

That, of course, has changed. Much of that material has been ‘digitised’ (transformed into a computerised format) which then became available, 24/7, 365 days of the year, via the internet.

Then ‘search engines’ were developed – things like Google, or specialised search engines like the National Library of Australia’s remarkably-useful Trove. These instantly identify history sources relating to whatever topic might be of interest: Google, for example (at the time of writing this) could direct us to 39,200 references to Aboriginal breastplates; Trove could identify 909 references to Aboriginal breastplates in Australian newspapers and gazettes alone.

Whilst both these (and other) search engines have made the process of researching history topics very much easier, they have not eliminated the need to perhaps read all the sources they have identified – in this example, up to 39,200 articles!

Which is why the recent development of A.I. has been so significant. Now A.I. can review all of those references and prepare a summary from them. For any researcher, this makes life very much easier, and researching so much more efficient. A.I., essentially, can be like having two, three, four or more research assistants simultaneously working on the one project.

The problem, though, is A.I. cannot always differentiate fact from fiction.

Misinformation (or mistakes) has always been a problem for history and historians. Unfortunately, once something (factually correct, or incorrect; or simply an historian’s opinion) is published it tends to become viewed as ‘fact’, and this situation has only been exacerbated by the increased use of media such as Facebook where anything (factually correct or total rubbish) can and is so easily published.

So, if A.I. cannot, or does not, ensure its resulting histories are factually accurate, over time ‘history’ may well become little more than collections of fiction!

Traditionally, historians have revealed the sources of what they are publishing. This, at least, enabled a reader to go back and verify the facts. With A.I., sources are not necessarily documented, so if A.I. picks up some ‘garbage fact’ and then presents it as ‘fact’, truth and accuracy become the casualties. (In A.I.-speak, these errors are called ‘hallucinations’!) And what’s to stop the radical Right or the looney Left developing A.I. ‘black boxes’ that simply churn out histories reflecting their own extreme points of view?

A.I., however, has many benefits apart from simply being able to process vast amounts of information. For a historian, it provides instant translations from other languages. It can also be asked to ‘fill in the gaps’ of damaged or illegible documents. And it can even efficiently do things that would take a human years to do (if they were successful at all!) like reassembling a badly smashed vase from its various small fragments…or even reconstituting entire ‘lost’ Aboriginal languages from multiple surviving sources that had each recorded just a little of the original language.

Traditional histories have also suffered from being based on partial evidence and cultural prejudices. A.I.-generated histories, in theory at least, should be more objective (so no longer will it be a case of ‘history is written by the victors’) and it is more likely to include material relating to the ‘peripheral, the obscure, the marginalised.’
And A.I. can provide a valuable ‘brainstorming’ facility to historians – suggesting new leads that might be explored, suggesting alternative ways of looking at events that happened in the past (the article on the Bermagui Mystery in Recollections 52 (available at www.southcoasthistory.org.au) presenting a good example of this).

The internet has certainly made history more accessible to the general community. The question now is whether A.I. will make it even more accessible and will make more people interested in history (by, for example, being able to provide a history on any specific topic of interest to an individual; go on – type into your search engine something esoteric like ‘write me an essay on the history of propellers’ and you’ll be instantly gratified!!!)), or whether increasing misinformation will ultimately consign history to the ‘dustbin of history’.