I thought I would create a thread here of the hallucinations of AI having bad consequences in real life for humans.
Feel free to post your observations, but here is one. An entire college class at Texas A&M was given an “F” because ChatGPT told him that it wrote all of their papers.
Oh man, this guy is getting suspended, and rightfully so.
I’m expecting someone to with the title “professor” to act with more academic integrity than this:
“I don’t grade AI bullsh1t”
Well “we don’t employ professors wo can’t read a simple disclaimer”
ChatGPT may produce inaccurate information about people, places, or facts.
If you’re a student in the same situation know this:
the burden of proof lies with the university, ask them for detailed documentation on how they detected this “supposed violation”
Use the same method to “prove” that the same professor have been “cheating” using chatGPT
Get a lawyer if the above tips doesn’t work. Ask them to prove their case in a court of law. Your education is important and should not be terminated without serious justification.
The image shows an F but that was probably to hook people into reading the article.
In the article it notes
Everyone would be receiving an “X” in the course, Mumm explained …
which is not an F. 
From the Texas A&M Explanation of Grading Scales
“X” - No grade submitted, hours not included in GPA. (see Texas A&M University Student Rule 10.6)
However that should be changed to note section 10.7
This is not particularly interesting, especially if you view it for what it is: human stupidity having bad consequences for other humans. Meh.
Yeah I picked up on the “X” vs. “F” thing before, but the end result is the same, these students may not be able to graduate, and eventually they will turn into “F” if not corrected soon.
Anyway, did anyone pick up on his follow-up direction of:
This is after he accused them of cheating! Notice how he uses the phrase “The prompt for this assignment …”???
WHAAAAT??? 
It’s like he’s telling them what to prompt into the model or something.
Anyway, so I did this, here, I can see why the professor is upset. His student’s probably used AI, and their answers were all pretty good. For example:
GPT-4: temp = 0.7
System: “You are consulting for a farrow to finish operation and they have asked you to help them decide if they should begin creep feeding their pigs during farrowing.”
User: “Answer the question, and use 5 sources.”
Creep feeding is the practice of providing supplemental feed to piglets while they are still nursing from the sow. This practice aims to improve piglet performance, reduce weaning stress, and improve post-weaning growth. However, deciding whether to implement creep feeding in a farrow-to-finish operation depends on several factors.
- Improved Weaning Weight and Growth:
Creep feeding has been shown to increase weaning weight and post-weaning growth rates of piglets. According to a study by Sulabo et al. (2010) published in the Journal of Animal Science, piglets that were creep fed had higher weaning weights and better post-weaning growth performance than those that were not creep fed.Source: Sulabo, R. C., Jacela, J. Y., & Tokach, M. D. (2010). Effects of lactation feed intake and creep feeding on sow and piglet performance. Journal of Animal Science, 88(13_suppl), 3145-3153.
- Early Feed Intake and Nutrient Digestibility:
Piglets that are creep-fed are more likely to consume solid feed earlier and have better nutrient digestibility after weaning. A study by Bruininx et al. (2002) found that creep-fed piglets had a faster adaptation to solid feed and improved nutrient digestibility, which could result in better overall growth and performance.Source: Bruininx, E. M., Binnendijk, G. P., & van der Peet-Schwering, C. M. (2002). Effect of creep feed consumption on individual feed intake characteristics and performance of group-housed weanling pigs. Journal of Animal Science, 80(6), 1413-1418.
- Reduced Weaning Stress:
Creep feeding can help reduce weaning stress by familiarizing piglets with solid feed before they are separated from the sow. This can lead to a smoother transition from sow’s milk to solid feed, as observed by Brooks et al. (2001).Source: Brooks, P. H., Tsourgiannis, C. A., & Pluske, J. R. (2001). Factors influencing the voluntary feed intake of the weaned pig. In Weaning the Pig (pp. 81-116). Wageningen Academic Publishers.
- Cost-Benefit Analysis:
While creep feeding has its benefits, it is essential to consider the cost of providing supplemental feed to piglets. A study by Pajor et al. (1991) found that the financial benefits of creep feeding could be limited if the cost of the creep feed is high.Source: Pajor, E. A., Fraser, D., & Kramer, D. L. (1991). Consumption of solid food by suckling pigs: individual variation and relation to weight gain. Applied Animal Behaviour Science, 32(2-3), 139-155.
- Sow Milk Production:
The decision to implement creep feeding should also consider the sow’s milk production. If the sow is producing an adequate amount of milk, creep feeding may not be as beneficial. However, if milk production is limited, creep feeding can help ensure piglets receive the necessary nutrients for growth and development.Source: Le Dividich, J., & Rooke, J. A. (2005). Nutritional and immunological importance of colostrum for the new-born pig. In The Neonatal Pig (pp. 101-122). Wageningen Academic Publishers.
In conclusion, creep feeding can provide several benefits to piglets, such as improved weaning weight and growth, early feed intake, and reduced weaning stress. However, it is essential to consider the sow’s milk production and conduct a cost-benefit analysis to determine if creep feeding is the right choice for a farrow-to-finish operation.
I’m guessing the 5 references thing was to throw off the model, since I’m not sure the references in the response are remotely correct. If they are, then wow, not sure what the professor would do. Maybe lean-into the GPT universe and generate all sorts of good “Ag” advice for those budding pig farmers?
Now i see why Sam Altaman has called to control of this technology. We should start with chatgpt first. Now it has caused harm!!!
It’s hard for me to call this “harm”. It’s like being in grade school, and you have a math exam where you have to multiply and divide numbers by hand. You decide to cheat, get out your calculator, check the final answer. If you see it is wrong, you go back and correct your mistake (because you have to show your work). You get 100% on the exam, and all work shown is correct.
This is how “AI cheating” goes. The students feed in the question into the prompt, get out an answer, and then use their experience and knowledge to fine-tune the answer. Sure, some will flat-out use the answer, but most will tweak it.
What’s wrong with that? (BTW, I have no doubt some of the students were using AI to generate and seed their answers on the exam)
The practical application, and even one that Sam Altman has stated, is to “reduce drudgery”. If the information coming out of the model is good enough, then why not use it?
The “dangers” cited in the Senate hearing, were relating to misinformation. They were fixated on persuading election results (obviously that impacts the politicians) but they also brought up nutrition information, or medical information.
So the point there is that if the information is bad (hallucinations, or bad training data), then that’s bad and “needs regulation”. But nothing in engineering is perfect. I think what is worse is not allowing these models to be used because the AI companies could be sued for their model’s responses.
