The Ai thread


On college campuses around America, old-fashioned blue books are making a comeback.

Around a century ago these standardized booklets for written exams were introduced by Butler University in Indianapolis. For many former students of a certain age — myself included — the sight of a blue book can generate anxiety and even nightmares. And many former instructors — myself again included — recall the tedium and strained eyesight of trying to decipher students’ handwriting. So it was an improvement when exam-taking shifted from paper to computer. Or so it seemed at the time.

Now, however, students are using AI to write essays and answer questions on take-home exams, as well as taking in-person exams on their computers. As a result, an AI-arms-race has developed between instructors and students. Concerned that students are not doing the work themselves but are simply copying and pasting AI output, instructors have begun using AI programs to detect students’ use of AI. Inevitably, there are now AI programs that students can use to outwit the instructors’ AI detection programs.

So, not surprisingly, many instructors are going back to handwritten in-class exams, generating a sudden boom in the demand for blue books. Ominously, even the return of in-person testing may not solve the problem of testing in the face of AI: Cheating using AI glasses is on the rise in Asia and will doubtless spread worldwide.
Don’t in-person computer-administered exams use Examsoft or something like it? Should prevent use of AI, at least.
 
Not to mention the fundamental problem of building the necessary expertise to be able to assess the validity of AI generated information.
The expert & AI combo is very powerful, but where do you get the future expert from?
 
A survey found Gen Z hates AI the most while also using it the most. How is that possible? They are being forced to use it at work.
 
On a side note, I just glanced at the Top500 list from November. I am not sure how well SCs correlate to LLM datacenters, but, adding up the demand for the 3 top SCs, the electricity used for one hour of time on all three adds up to enough to run my car for around three hundred thousand miles (more, probably, but my efficiency drops for 3 or 4 months in winter). If power usage for "AI" datacenters is at all like this, we clearly need to re-evaluate our priorities.
 
Wondering who people are using for their preferred LLM right now? The only one I pay for is Claude and that's mostly because I like to code with it. ChatGPT has gotten so limited on their free tier that I only get a couple of questions in before my time is up so I can't even really gauge whether or not it's worth paying for. I sometimes toggle between Gemini and ChatGPT though while Claude is churning away on heavier tasks.
 
Wondering who people are using for their preferred LLM right now? The only one I pay for is Claude and that's mostly because I like to code with it. ChatGPT has gotten so limited on their free tier that I only get a couple of questions in before my time is up so I can't even really gauge whether or not it's worth paying for. I sometimes toggle between Gemini and ChatGPT though while Claude is churning away on heavier tasks.
For me, it is between Gemini, I have pro from work, and it is included in my Google plan for my personal use. Ran into some token limits on using the Pro model a few times. Also, I use NotebookLM (again, I get to use it for free from work) for schoolwork. Also, I paid for Grammarly for schoolwork. The schoolwork use is that I am going for my Master's (at my advanced age), and I have a learning disability. NotebookLM takes all the Harvard Business Reviews and turns them into a podcast I listen to before I try to read that dense, boring material. I use Grammarly because of dyslexia and my atrocious grammar. Oh, and I always have to cite them in any papers I write.
 


A different take on the AI bubble (the poster above is sharing it, didn’t write it) - that it isn’t actually a bubble like the dotcom bubble was where private equity was underwriting the market. Here the author alleges that it’s public spending by the government that is backstopping everything which makes it less precarious in some ways but more so in others. Basically, if the US government retreats in its spending due to concerns about its deficit, that would trigger a crash more so than private investors getting jittery.
 
Wondering who people are using for their preferred LLM right now? The only one I pay for is Claude and that's mostly because I like to code with it. ChatGPT has gotten so limited on their free tier that I only get a couple of questions in before my time is up so I can't even really gauge whether or not it's worth paying for. I sometimes toggle between Gemini and ChatGPT though while Claude is churning away on heavier tasks.


I haven't used any of these services and I don't know where it is comparably at now, but my perception is they tossed out some free versions in the beginning and then very quickly transitioned to paid tiers if you want it to be worth a damn, but I could be wrong. If true, I don't think they had the functional free versions out there long enough to get enough people addicted to the point that they suddenly felt the need to pay for it. It's coming down to people who see it as a way to help improve their income, not casual users like you have with social media and streaming services.

Regardless, from all the people I've seen crunching the numbers from the outside there's no way based on expenditures that they'll be able to turn a profit from individual users, especially as subscription prices for everything are going up and people are deciding which they can live without.
 
All it means is that they'll be more stealthy about it. A couple hundred here and there. More gradual "re-sikilling" of personnel.
Nah truthfully they were overselling their product to investors but in a way that pissed everyone off. Don’t get me wrong people lost their jobs (or didn’t get hired, especially new hires) because idiot CEOs believed them, but the apocalyptic job claims were in the same vein as their models being too dangerous to release or even sentient. It was marketing to sociopathic investors who would hear that and think “I’ll make even more money!” Short term thinking to win the investors while alienating everyone else.
 
Nah truthfully they were overselling their product to investors but in a way that pissed everyone off. Don’t get me wrong people lost their jobs (or didn’t get hired, especially new hires) because idiot CEOs believed them, but the apocalyptic job claims were in the same vein as their models being too dangerous to release or even sentient. It was marketing to sociopathic investors who would hear that and think “I’ll make even more money!” Short term thinking to win the investors while alienating everyone else.
Well, agree to disagree. I've seen it in action. They've created models that are replacing people very handily. It's not going to stop - CEO's and CFO's get massive bonuses for their budget cuts....there's always going to continue to be pressure in a race to the bottom.
 
Using Google less and less because I know anything I type in will turn into ads. So far this is not the case for ChatGPT or Claude and for quick basic searches I've just been using DuckDuckGo.
 
Back
Top