Diatribes of Jay

This is a blog of essays on public policy. It shuns ideology and applies facts, logic and math to economic, social and political problems. It has a subject-matter index, a list of recent posts, and permalinks at the ends of posts. Comments are moderated and may take time to appear. Note: Profile updated 4/7/12

24 September 2023

My Kafkaesque-Orwellian Encounter with AI

I just had an enounter with AI so bizarre that I feel I have to report it immediately. It suggests that AI has the potential to turn what we conceive as fact and reality into a contingent and ever-changing phantom. It hints that AI could be used as a tool to gaslight anyone from selected individuals like me to our entire species.

On the other hand, what I encountered could have been just a very sophisticated example of Chinese hacking. Or it could have been a bizarre blunder of a harried programmer, of the kind that observant Web users see every day. You’ll have to decide for yourself.

The encounter involved Microsoft’s new AI-enhanced search engine Bing. In its favor, I find it more powerful than Google. It often gets me quickly to what I want, but sometimes at the cost of having to wade through wildly irrelevant and even bizarre search results.

But this time, the bizarre and scary search result, which I at first thought was right on, came up first. And it involved my own Google Blooger posts, which I know well.

This first search-result link led to an essay that I never wrote. Instead, it was an essay that combined a title of one of my essays with the body (the text) of another.

The chimera looked exactly like a real product of Google’s Blogger, on which I had written both essays years before. But it wasn’t my work; it was a hybrid.

Worse yet, the hybrid appeared designed to make China look good. Even still worse yet, the link to the false hybrid essay has the logical effect of hiding my real first essay from everyone looking for it using the search engine Bing. For those searchers, my first essay would not really exist, and the second bogus hybrid essay that was linked would have an incongruous title that doesn’t fit its text.

Since my first essay was unfavorable not only to China, but to its entire ancient system of writing, and the second essay was a flight of future-historical speculation favorable to China (before Xi Jinping went all despotic and removed his term limit as supreme leader), it’s possible to make a credible inference of sophisticated Chinese hacking.

To preserve the evidence, and to make sure I wasn’t hallucinating, I copied the key pages in PDF format, after making a copyright-infringement take-down request to Bing. I couldn’t find a way to upload those PDF files to this blog legibly, but here’s what the first Bing search result looked like:
When I clicked on the first bold link, labelled Diatribes of Jay; China Rising II; The Hantsu Hypothesis, what I got was this. If you scrutize this result carefully, you’ll see it’s a combination of the title from this essay, China Rising II: The Hantsu Hypothesis, and the body, but not the title, of this essay: China Rising III: American and China in Crisis.

So searchers using Bing to find the first essay would find only the second, albeit with the wrong title. They would never know that the first essay ever existed, let alone that it contained a lengthy scholarly exploration of the practical and human consequnces of China’s ancient system of writing. All they would know, if they were particularly attentive, was that the title didn’t match the substance of the essay they were reading, which would be one speculating on a favorable future history of China.

What’s even more bizarre is the dates in the URLs, which you can read in the search fields of your browser. The URL month-date of the real first (hantsu) essay is 2009/01. The URL month-date of the hybrid, bogus one is 2009/02, just like the real date of the second essay. But I reviewed the archived chronological records of my blog and found no record of a hantsu blog in February. How some programmer accomplished all this is beyond my comprehension. If deliberate, it was likely an inside job.

Orwellian? Kafkaesque? Scary? You bet! Choose your adjective. Now imagine that the subject matter were not just the musings of an aging, retired professorial blogger, but the words or summaries of statues, the opinions of a court (maybe the Supreme Court), or official records of an event like the Capitol Insurrection of January 6, 2021.

If nothing else, I hope that this brief, true tale evokes as much justified suspicion of and concern about AI in everyone reading it as it did in me.

For brief descriptions of and links to recent posts, click here. For an inverse-chronological list with links to all posts after January 23, 2017, click here. For a subject-matter index to posts before that date, click here.

Permalink to this post