"Jerk" it is not. The perils of AI.
"Jerk" it is not. The perils of AI.
While the general public seems mostly oblivious to the weaknesses of AI and takes it at face value as "intelligent", in our industry we see the shortcomings everyday even while we are trying to find ways of leveraging it to make our work better. At this point, the evidence is clear that we should be very skeptical of anything produced by AI and that the advantages it provides may be more than cancelled out by the quality and shallowness of the results. For many reasons. Here's one:
Source: Det krävs varken att du är partisk eller agerar på ett visst sätt för att du ska anses jävig.
AI: It is neither required that you are biased nor act in a certain way for you to be considered a jerk.
For my non-Swedish readers, the source text does not say anything about "jerk". Instead the concept is "conflict of interest". But that didn't fit the algorithm's recognized pattern so it went with what statistically made more sense, jerk. There was no thought, no intelligence, no awareness in what it offered up. Because there is no intelligence in AI, only the appearance and assumption of intelligence. Much like any second-year student (sophomore) who has decided they've figured this academic thing out and can use big words to sound smart without understanding them. While this statement from an AI engine may very well be true on a societal level, the error is a fatal flaw in a translation. And a less than thorough review of the text could allow something like this to slip through to the end client.
In my private reading, I am confronted daily with texts that are clearly AI driven. Many I suspect have never had a human proofer or editor. These are in online publications that pour over us like molasses. And the more I read them, the less respect I have for them, and not just for humorous mistakes like the one above.
If we have any intellectual abilities left, we are both consciously and unconsciously filtering these texts out very quickly. The problem is how much time it takes to determine they are AI-written texts. The question of plausibility comes into play. Usually we have to invest a bit of time before we begin seeing the lack of logic and flow and the sophomorish writing style in these texts, where they build up to a conclusion that is never given. Repeating the same facts (or false fact) with slightly different formulations seems to be a valid writing style here while never actually coming to a point.
The question is how I can eliminate them from my life feed. And how do I access the useful bits of Facebook (groups) and other social media, without being confronted with all the AI junk? I've not yet found an answer to this. It may require simply eliminating all non-trusted sources from my newsfeed. Easier said than done.
Update: I’ve solved it by deleting Facebook and most other social media. It is a cesspool of manipulation that I’m no longer able to defend against. Better to live life than throw away any more time in it.