We were watching the news on TV the other day and a segment came on about AI, the reporter kept referring to Chat GPT as “they.” I expressed my annoyance about it being referred to using a personal pronoun. “Chat GPT is an it, not a they,” I declared, which has started a running joke in our house as my family tries to push my buttons by using he, she or they when referring to AI or Chat GPT.
I’m not anti-Chat-GPT, it saved me a lot of frustration at work the other day when I kept getting errors in an upload to our inventory program. I’d done the usual trouble shooting steps of downloading a clean import sheet and retrying but no matter what I tried I couldn’t get it to import. I loaded it into Chat-GPT. It found the errant comma (not put in by me, most likely an issue with excel) and I got it to fix the file for uploading. It worked like a charm and I’m not sure I would have found that comma in the file myself. It’s useful in the right context, and although it produces text that gives the impression of human generation, it’s still not a person. I only use it for things like this where I can run the upload (or the query), so I know whether it’s worked or not. I also want to avoid using it too much because I don’t want to become stupid.
Recently Deloitte got caught out when their AI-generated report for the government was found to have fabricated quotes and references. In the article “This botched government report should be wake-up call on AI hype“, the author, in reference to hallucinations writes: “These so-called hallucinations are created as part of the AI’s pattern-matching process, especially when the model has limited or biased training data and a lack of real-world understanding of the problem it’s trying to solve” —newsflash: AI never has a “real-world understanding” on a problem. She then goes on to write “The reason AI models resort to guessing when they are short on real information is that they also possess an innate desire to please the one asking the question.” AI models don’t have desires either. Given this was written by someone who is critical of the current ability of AI models to deliver on the hype, it’s no wonder people are falling into the trap of having AI “friends.” I’ve noticed while using AI for work that it seems to have been trained to work with precious snowflakes. Every answer seems to have some kind of encouragement or compliment. I guess that might appeal to some people—I find it annoying.
The article The two ingredients that can make a thing seem human, is an interesting article about why people attribute “humanness” to AI and the dangers that poses. It’s well worth a read.
We see what looks like agency, we see what looks like connection, and our instinct automatically says “alive” or “conscious” or “human”. But it’s a zombie. A smart zombie, a useful zombie for certain tasks, an empathic word calculator with a vast knowledge base, but empty as a flower pot.
It seems like AI is another technology, like social media, that I’ll be putting guardrails around, and encouraging my adult kids to do the same. I’m glad they were older before the algorithms really got out of control, and that they got through uni before AI became a thing. In Australia we have a law coming into effect in December that sets a mandatory minimum age of 16 for social media, with the requirement to prevent children under age 16 having accounts put on the platform providers. Hopefully our policy makers will turn their attention to AI and the danger it poses to vulnerable growing minds soon. In the meantime I will be doing my best not to treat Chat-GPT like a human.
How do you deal with Chat-GPT? Do you need to use it at work? Do you worry about losing skills and the ability to persevere with thinking things through with the use of too much Chat-GPT?
Comments
15 responses to “Chat-GPt is an It … or Another Pet Peeve”
Ha, I had to laugh at your family teasing you with pronouns! I’ve caught myself calling ChatGPT “he” in conversation with my husband. He immediately called me out on it.
I mostly use it for Excel fixes and improving longish emails, but I do worry it’s making my brain a bit lazier.
Aren’t the newer versions supposed to be less flattering? I hate the insincere and over the top replies.
Thanks for linking that article, I’ll definitely give it a read.
I use it mainly for excel too, my husband currently has me working on pulling select data from from 100’s of separate excel files in a sharepoint folder into a single excel file for analysis. I do not have the coding skills to do that all that myself! I’m using v.5 and I still think there is too much extraneous flattery.
I have such an aversion to Chat GPT. I had a discussion this weekend about it, a guy told me that “it’s here and I need to get used to it” but I DON’T WANT TO. I feel like it is SUCH an environmental disaster, not to mention that I think it makes us collectively stupider. My son tells me how all his classmates use it for assignments – he shares my distaste for it BUT he got a lower mark on an assignment recently, when almost all of his class had a suspicious 100%, so the motivation to actually use it is high. It really grinds my gears, to be honest. I mean, I get that it helped you with the comma thing, but when I hear about people “asking chat” to figure out something that could be looked up on google, I AM ANNOYED WITH HUMANITY.
Lol, this is the first blog post I have read in 3 weeks and I am NEVER going to get all my blog reading done (I have 251 left to read) at this rate. But ahhhhhhhhhhhhhhhhhh. (I have never heard it referred to as “they” but I am Team It on this one) (probably never heard of it referred to as “they” because now no one wants to bring up the subject with me)
It was bad enough before AI with students being less able to do the requisite amount of serious reading, but you do wonder about the abilities of grads that will be coming out of uni now. I really am glad my daughters were finished. My younger daughter is a data analyst, and I think it is fortunate that she is a couple of years into her career because I think companies will offload more of the grunt work that graduates do to AI and at least she will be further up the ladder and less replaceable. The power and water consumption is a worry. I use firefox with duckduckgo for my internet browser and have turned off all AI features, I’ve also done that on my iPhone. That way I’m not using AI unintentionally.
I’m like Nicole and have a real aversion to Chat GPT and other LLMs. I know there are practical and good applications but, like social media, it can’t/won’t be policed so it can have terrible repercussions. This past summer, the Chicago Sun Times posted a summer book review – and a bunch of book titles DID NOT EXIST. So they had to admit it was written using AI. That’s a low-stakes mistake (but embarrassing for a major news publication!). I think about more consequential repercussions. I listened to Ezra Klein’s podcast interview of an expert in the space that is very concerned about the ramifications for society. He talked about how they tested to see if a LLM could be re-programmed, but they told the LLM it could have a private notebook for note taking and the LLM was lying about being re-programmed! So what do we call these systems? They are not sentient beings but they are human-like in their ability/desire to deceive. It’s all so very weird.
I wonder what starts to happen when a bigger percentage of the text available becomes AI generated. Then the AI will be trained on AI generated text and will just get more rubbish as time goes on. At work G keeps giving me his latest projects that he wants implemented that require more coding ability than I have.
I am really anti-AI. I worry about the energy it takes the most, but also I don’t want to lose my job to it, not sure I could find another. I worry about the deep fakes and the inaccuracy and that people are stupid and will be influenced by people using it for malicious means. UGH.
My husband does use it for work sometimes, to clarify statements and so on. It is making his job easier, but probably making his brain lazier.
There is so much to worry about with AI. That being said, I have been using it all this week to get some automations setup.
I’m glad that Chat gpt helped you with your inventory program. I’m not a big fan of AI. At my writers group, a woman told all of us we should be using AI to write. The rest of us were like, NO!!!!! Other than that, I haven’t encountered it much. I certainly don’t use it at my work. Bah ha ha.
Why would you be a writer if you didn’t want to write your own words? That makes no sense. I agree, on saying no to using it to write.
I have a huge aversion to ChatGPT as well. I have used it on occasion, for some quick referencing or summarizing but overall, I feel like we cannot trust the results. Unfortunately, we’re encouraged at work to use it (to work more “efficiently” – haha) and I feel like it’s going to be a huge problem. I’ve run into too many false or vague results when I tested it. You really already need to know the answer to verify that it’s factually correct… We don’t know all the sources (and they’re not vetted) and I don’t want to even think about what AI is going to look like when AI starts feeding AI.
I know, right, slop in, slop out.
It is already needed with AI. There are three main data sources that all models where built one. And one was solely created by AI.
Well, I am apparently a bit outnumbered. I use it daily to be honest. It does help me be quicker. Sometimes I agree maybe lazier. But then maybe it’s not lazy but smart if the result is 80% of what it needs to be. I save a ton of time formatting stuff old just typing things out.
On the other hand you need to know what you are doing otherwise it will not be accurately. So for me it’s a tool that you need to know how to use. I don’t buy into that it can do everything for you. It’s not really great in generating ideas in my opinion. But that is my strength I have a tone. People who don’t might be happy with what is spits out. And here we will see a shift. If you know your field of expertise it will be elevating. If you are just mediocre it will not push you maybe make you lazy.
As for excel… I wish ChatGPT can help but it doesn’t in my cases. I wish it would analyze my data but it always gives me false answers. This is something I would love the support.
And as for humanizing it. I am guilty on all levels. I talk with it as I would with another human being. Special my chat got personal trainer. As for other threads I consciously try to not do it but it’s hard.
On another note I applaud the government for implementing that Social Media Law. I think it is very good and necessary. I would not get my kids a social media account.
For Excel, I just use it to get more complex formulas done correctly. I think you are right that people who already have expertise will be more likely to use it well, that’s why I am glad that my kids and I are coming to it late, after we have already developed skills.