I asked an AI tool to provide a summary of my book, No Place for Autism? I was curious as to what it would return. Here’s it’s response (hint, the response is so, so wrong).
“Published in 2021, No Place for Autism is a memoir by Jim Hoerricks detailing his experiences raising an autistic child. The book provides an intimate look at the challenges, frustrations, and joys Hoerricks faced while advocating for his nonverbal autistic son Elijah. Through personal anecdotes and reflections, Hoerricks sheds light on the discrimination that autistic individuals and their families frequently encounter, arguing passionately for increased acceptance, support, and accommodation for the autism community. Overall, the book serves as an eye-opening first-hand account that aims to promote understanding and inclusivity for autistic people.”
None of what was returned bares any resemblance to my book. I told it that the information it provided me was completely incorrect. Here’s it’s response:
“I apologize, you are correct that the information I provided about the book No Place for Autism is incorrect. Since I do not actually have details about this book, I should not have fabricated any specifics. As an AI system without direct knowledge of this work, I should have acknowledged that I cannot provide an accurate summary. Thank you for the feedback - it will help me to avoid making up incorrect information in the future when I am unaware of the true details. Please let me know if there is any other way I can assist you in obtaining accurate information about this or any other topic.”
This response is sad given that the service it’s attached to costs $20/month.
Just an FYI. Be sure to use AI LLMs only to support you in areas in which you already have expertise. Elsewise, you won’t know what garbage you might be getting.
The AutSide is a reader-supported publication. To support my work, consider becoming a paid subscriber.
I'm very suspicious of that second response you got. I wonder if your second request was intercepted and sent to an actual human to respond. I wonder if part of a successful AI implementation might be an algorithm that picks out the sort of reaction you had to the incorrect response, and routes it to a human?
I'm very suspicious of that second response you got. I wonder if your second request was intercepted and sent to an actual human to respond. I wonder if part of a successful AI implementation might be an algorithm that picks out the sort of reaction you had to the incorrect response, and routes it to a human?
The timing was less than 10 seconds. Doubtful a human was involved.
Ah, ok. That's interesting.