I think you are right, AI is not at the stage where it could take over the job of carrying out and updating systematic reviews. But I also agree with your conclusion that AI-assisted systematic reviewing could have a positive contribution to make.
There are tools that can classify huge corpora into topics, that can help work out a comprehensive taxonomy, that can extract metadata, that can anonymize data, that can summarize data, that can answer questions via a chatbot, that can spot repetition and plagiarism, that can give diagnoses in response to a list of symptoms, etc. Of course, human input is needed, and will probably always be needed in a lot of critical exercises.
As for the anecdote, if the algorithms were the problem then the solution should have come from the software vendor. If the data was the problem then the editor should have got his data experts to analyze it to find out why it was doing unexpected things.
If supervised machine learning algorithms are 'trained' using biased data, they will learn the same biases. But 'unsupervised' or 'semi-supervised' algorithms work out topics by clustering and other processes, so they may have less of a tendency towards biases.
Then, journal editors may allow an unsupervised algorithm to select reviewers and be surprised that some of the reviewers selected are, for example, those with no academic affiliation. (Some journals reject those with no academic affiliation, which could lead to biases or compound existing biases.)
A different selection process might be able to select reviewers who are perfectly capable, perhaps even more capable than ones the editor might have selected.
Of course, this is all speculative! But some interesting and unexpected results may come from the enormous amounts of AI research that is currently taking place.
HIFA profile: Simon Collery is a Freelance Data and Information Analyst, based in London. He has worked as a non-profit manager in Tanzania, Cambodia and Kenya. He studies and writes about HIV, particularly non-sexually transmitted HIV, via unsafe healthcare, cosmetic and other practices.