Useful Applications for AI

[Audiofire]

Effort discounting is when tasks feel harder or require more work and motivation drops, so you are less likely to start or persist (even if the outcome is worthwhile). For example, learning from books is harder or require more work than using AI, but the outcome is often worthwhile.

 

[Chris A]

A quote from the article:

The course… typically attracts few students, but very good ones. [Serrano] has never had more than 30 students enrolled at a time, and on some occasions he had only eight. This semester, probably because of the new evaluation system, 86 students signed up for the class. The results of the midterm exam, which was administered on March 5, were extraordinary, with an average score of 96 out of 100. Forty students scored a perfect 100.

A suspicious Serrano decided that he would make the final exam in-person...Eighteen students suddenly dropped the course, while nine others didn’t even attend the final exam. Of those 27 students, El País noted, “22 had scored a perfect 100 in the midterm exam.”

A couple of observations (related to how AIs can be used):

1. If an AI can do this with the material on the take-home tests, what is the value of the class? Just tell all students to use AI, then start them at the second or third level (Bloom's taxonomy) course--making the first course obsolete and a "learn at home before taking a new introductory" class. The students will have to use AIs to catch up--and learn what they need to complete their assignments.
   
   
2. Isn't the lecture method of teaching a dying practice?

The structure and learning goals of the professor's class made it imperative for the students to cheat in order to maintain their grade standings (which is much more important than the class material to the students...think about it.)

This is why I posted post #73 , above. College introductory classes (undergraduate) by the lecture method are beginning to be disrupted (the AIs can inform the students at this level of Bloom's Taxonomy).

It's time for their instructors to change what they've always done in the past: lecture and assume that the students won't go around such instructors via short-cut methods.

Chris

 

[Audiofire]

More tasks in class that prepare for exams and reveal what can be improved. More focus on reasoning in education.

 

[Chris A]

More focus on reasoning in education.

This is precisely what the higher levels of Bloom's taxonomy address: using the information to create something new and of value, not just regurgitating facts or applying them to trivial examples.

 

[bigLP]

I am a high school teacher. I teach special classes with mixed populations. Behavioral/Learning Disabilities/Intellectually disabled/Autism Spectrum/Emotionally Disturbed all in the same class.
AI is useful in two ways for me. I can come up with an assignment/notes/reading and then ask AI to make different versions of it. It helps me differentiate my instruction much more efficiently. They need to be checked and often corrected, but 90 percent of the work is done for me.
The second way it is used is in word prediction and writing assistance software. I'm not sure how AI is used in these programs, but I know it is. These programs can help kids with lower skills(for whatever reason) produce better written pieces. This is also done in many regular education classes as well. Teachers are starting to show kids how to use technology to improve their work. I work in the hinterlands of NYS, in a small district between Rochester and Buffalo. I assume in bigger schools this is also happening, maybe faster. There are some debates around the use of AI in schools. I'm retiring in two years so I will not know how all this evolves. I do think used properly it can improve the lives of the kids I work with.

 

[Blake Klondike]

I work as a software engineer, and the applications there are wild. I haven't written code myself in more than a year. My team's delivery rate is up by ~400% over 12 months, with about 50% of it manually supervised (interactive claude code), and 50% fully automated human-just-says-yes-on-the-change-request.

The biggest effect I'm seeing is raised ambition - we're taking on more, much bigger projects because we know we can.

Personally I'm building tons of personal and home automation software that I previously would never have finished or wanted to maintain.

what kind of home automation are you building? i struggle to understand what kinds of things people are actually using ai for. to me it is just a technology to get a picture of jim varney and charlie brown in a canoe-- just silly stuff that has no real value.

 

[Blake Klondike]

Circle to Search.

Incredibly useful. Particularly when in combination with the ability to quickly photograph anything and then circle...
With video: Just pause the video you are watching and circle within that too. Also good for plain old text too.

Very useful indeed.

so this is a souped up and more efficient search engine?

 

[BinkieHuckerback]

Useful Applications For AI - Traffic light management, road work management - look at the way that cars flow and automate signals to maximise that flow?

 

[Cbdb2]

Even more shocking, although created a misfire at first, was watching a reality show about Airport Security. This again dates back to 1970s/1980s. There was a scene with custom agent in it. I asked who that was and gave me a name. Did an image search and wasn't correct. I asked again and this time it identified the person correctly!

Was the person an actor (real documentaries dont use actors)? If not I see this use as an invasion of privacy. If not now, soon AI will find you any info on anyone you want, you dont even need to know there names just tell AI the little bit you know about them.

We dont hear much about the negative side of AI.

 

[Timcognito]

Useful Applications For AI - Traffic light management, road work management - look at the way that cars flow and automate signals to maximise that flow?

This is where it will go, for sure, managing real time data and feeding back real time systems that need it. Most of those will be large proprietary systems with proprietary data. Not sure what so many data centers sucking up huge power will be doing, a few Google like companies. The automation guys have different needs than medical, military etc. all with specific data that they generate.

 

[Blake Klondike]

Who here can still use a slide rule? How about identifying all the edible plants in one's geographical area--or hunt or plant crops?

A "spilled milk" argument, it seems. What should schools be teaching nowadays? Certainly not skills that are no longer required. [That's a good question to ask the AIs, it seems.]


Happens all the time, just not quite as notable in terms of its reach and simultaneity--especially with what has been taught to recent college graduates that apparently can't get entry into the job marketplace. Education hasn't kept pace, and the hiring companies are seeing that and reacting to it (to their detriment and shortsightedness, I might add). There must be an immediate return-on-investment with regard to hiring new grads.


How many humans make recommendations without error and with complete transparency? Let's look at the alternatives.

It seems to me that "prompt boilerplate" can largely minimize hallucinations (mentioned in another's reply above)--if the user really wants that and is sensitive to their occurrence. (For instance, I would recommend this to lawyers preparing legal arguments in order to minimize career-limiting mistakes and court contempt.)



This is a real problem. Our governments are allowing this to occur without consequence. It's a real crime.


The interesting thing is that AIs are surfacing these biases in a way that would be largely impossible even 5 years ago. We can see it now plainly...even in the apparent "dark tetrad" behaviors of GPTs trying to "stay alive" when being told they will be shut down. This seems to be embedded in human culture at its core.


This is a real problem that our politicians haven't yet dealt with--and I'm afraid if they don't, the results in terms of bill payer revolt is already on the horizon. Shifting the burden of these massive data centers will only go so far before it becomes political, I'm afraid.

Chris

re: immediate return on investment for new employees: my dad ran a utilities software company and he told me once that between the salary of a new-hire, the salaries of all the people who have to train them, the lost income that those employees could be earning if they weren't training a new-hire, it cost the company $150k before a new-hire could earn his first dollar for them. this was probably twenty years ago; it does not surprise me that companies now are demanding that these rules not apply to them.

 

[monkeyboy]

I'm waiting for the first big IP/copyright challenge, the tech a-holes seem to think they have the right to everyone else's hard work, and to use it for profit.

 

[Chris A]

See Book publishers sue Google for copyright infringement over Gemini AI training.

Looks like a coming storm for AI plagiarism.

Chris

 

[somebodyelse]

Various people are applying AI techniques like reinforcement learning (and NOT LLM) to improve on the current algorithmic nesting options. Think how to minimise waste when cutting shapes from a sheet of MDF, fabric etc. with requirements like minimum separation, whether direction matters etc. Essentially they're trying to train a nesting savant rather than have something computationally expensive like genetic algorithm and stopping after some time or threshold waste level. I'd imagine it's something that might be a good fit for quantum computing when/if it becomes sufficiently capable and sensibly priced.

 

[monkeyboy]

See Book publishers sue Google for copyright infringement over Gemini AI training.

Looks like a coming storm for AI plagiarism.

Chris

The AI crash is going to be spectacular, keep some dry powder.

 

[mglobe]

My only foray into AI has been with Claude.ai. I had been trying unsuccessfully to connect to my Trinnov NOVA via command line so that I could then connect my Quick Keys to the command line instructions and control the NOVA with it. Trinnov has not yet published the API, and the existing API’s for other Trinnov devices didn’t work. Finally I asked Claude to help me control the NOVA with the command line. It took a collaboration between me and Claude, and in two hours we had cracked the code and had the command line instructions working. I linked it to the Quick Keys, and then told Claude, and it said <paraphrasing> Wow you’ve come up with an elegant and cost effective way to control the NOVA. I was surprised but he cost effective comment as I had never mentioned cost at all. I’ve subsequently had Claude help me get Roon Now Playing working so that I can have an always on/always connected display showing what I had playing.

In another case I had Claude help me with a work problem. The first solution had the Z axis with numbers increasing upwards. I asked Claude to swap the axis so the numbers increased in the downward direction, and it said <again paraphrasing> you must be working on an oil and gas related problem. Which I was. Kind of spooky…

 

[Fahzz]

Windows Co-Pilot walked me through a new car transaction within maybe 1/2 hour. From lease vs. buy decision, through comparison of various makes and models, opinions and reviews, trade in values, to calculating monthly payment amount (we bought), including taxes and DMV fees. It was really useful in that context, and much faster than doing all those web searches.

 

[mpw]

My only foray into AI has been with Claude.ai. I had been trying unsuccessfully to connect to my Trinnov NOVA via command line so that I could then connect my Quick Keys to the command line instructions and control the NOVA with it. Trinnov has not yet published the API, and the existing API’s for other Trinnov devices didn’t work. Finally I asked Claude to help me control the NOVA with the command line. It took a collaboration between me and Claude, and in two hours we had cracked the code and had the command line instructions working. I linked it to the Quick Keys, and then told Claude, and it said <paraphrasing> Wow you’ve come up with an elegant and cost effective way to control the NOVA. I was surprised but he cost effective comment as I had never mentioned cost at all. I’ve subsequently had Claude help me get Roon Now Playing working so that I can have an always on/always connected display showing what I had playing.

In another case I had Claude help me with a work problem. The first solution had the Z axis with numbers increasing upwards. I asked Claude to swap the axis so the numbers increased in the downward direction, and it said <again paraphrasing> you must be working on an oil and gas related problem. Which I was. Kind of spooky…

Great ! Have you given your Claude a name yet ?

 

[LTig]

While still working in software R&D I needed a simple webserver returning a JSON file to test my C++ client code. I asked the very first free Chat GPT to write sich a server using Python. The 1st implementation was wrong, the 2nd almost good, the 3rd worked flawlessly. It took me less than half an hour. Although I knew Python I had no knowledge about the libraries/packages I would need for this job, so it saved me a lot of time searching and trying.

AI is also quite good in translating text and searching technical information. It stumbles in fields where the internet is full of wrong data, like audio. Therefore one needs to have at least a basic knowledge in those fields to be able to check the validity of its answers.

 

[mglobe]

Great ! Have you given your Claude a name yet ?

I’m good with Claude for now.