How do you use AI?

I was a non-believer with the productivity gains and avoided using ChatGPT. Fast forward some time, and as I'm chartering my group's path forward with big data and AI for R&D, I've setup a server with Ollama (a tool to load opensource AI models) and am finding myself using it a lot for all sorts of things; from the typical Python aid, trying to circumvent learning SQL syntax, bypassing Google for digging deeper into recent blood labs, to creating kids jokes at dinner, etc etc.

Curious how others here use AI.

Oh, and yeah, I saw somebody got good with Stable Diffusion and is making avatars for forum members. Definitely a cool idea.

 

I use chatgpt a lot. In fact I just asked the sample question below:


Prompt:
A guy on a truck forum is asking how people are using you for truck-related stuff. Do you get a lot of queries on toyota tundra trucks?


Completion:
I do get queries about trucks, including Toyota Tundra models, but the topics can vary widely. Common truck-related questions include:

• Recommendations for mods and upgrades (e.g., lift kits, suspension, and tires)
• Troubleshooting issues (engine problems, electrical systems, etc.)
• Towing capacity and vehicle specs
• Camper setups, like turning trucks into overland or RV vehicles
• Battery and electrical system configurations for off-grid use
• Fuel efficiency tips for heavy-duty use

The Toyota Tundra often comes up in discussions around towing, reliability, and off-road capability. There's also a lot of interest in outfitting it for camping, off-roading, or as a daily driver. Do you have a Tundra, or are you curious about certain features?

 

When the terminators come I’m pointing them y’all’s way.

 

I'm hoping it becomes self aware soon, and we can vote for it to be President.

 

Actually, just curious about any of the ways you find it useful. Was hoping to get ideas from others since application/implementation isn't always intuitive. It gets news coverage for the impending job replacement for coding all the time, as well as RAG and autonomy, but there surely are other, better every-day use cases. As an example, saw a YT video talking about the project 'Fabric' where the author's goal was to essentially digitize and database his work and personal notes to offload his memory demand. https://github.com/danielmiessler/fabric

 

My day job is in education as the Director of Digital Teaching and Learning for a K-12 school system. We have *obviously* been tracking Generative AI and its impacts on our classrooms and education in general for a couple years. I have a lot to say about the topic, but to answer the question from this thread: I use it quite a bit in my personal and professional life. Having the ability to upload anonymized data and documents and receive quick analyses of such has been a massive timesaver. Example: 900+ responses to a survey. Spreadsheet uploaded to LLM and prompted to review responses and triage the needed support areas. Had a suggested attention list back in 20 seconds as opposed to several hours of manual data aggregation.

 

@theTupper - expand a bit, if you don't mind. I have young kids starting school and would like your perspective. My current thought is to block outside AI sources and tweak self-hosted models to be kid-specific as they get to the age/grade that it becomes part of curriculum. Also, are you using a private account for Chat that it's allowed to upload data or do you self-host?

 

Great questions.

Firstly—we don't block any of the AI tools in district. Students under 13 aren't allowed to use the big players' tools due to COPPA, so any exposure IN THE CLASSROOM comes as teacher-led activities. The real objective during this time is literacy; making sure students understand the inherent risks and downfalls (misinformation, bias, etc) that are present with the benefits of GenAI. I don't believe that childen in primary/elementary school should be left to use GenAI tools without guidance, even if COPPA didn't outright forbid it. I'm personally not a believer in blocking what kids will use anyway (and that they will arguably need in the future anyway), and instead believe that education, literacy, and understanding is paramount.

Second—I perhaps used the word 'data' a bit too flippantly. I never include PII or any kind of data that is critical in any AI tool usage. That said, in the example above, I used ChatGPT. We don't have servers available to privately host AI models.

 

In my day job I used it a log to write simple PowerShell scripts, and that's about it for now. LOL

 

The company I work for has an app that uses AI to evaluate the health of roads with pictures taken from an iPhone. Different that what other people are talking about, but still neato. And don't ask me details about how it works, I have zero clue.

 

It has become integral to my work. I use it to rough out python programming, do historical research of sites, help find trends in large datasets or old reports, etc etc. it saves a huge amount of time.

 

As some others said, I use it to help hash out some codes that could take DAYS if I am not careful and if you code you know what kinda hell that is....
I also use it to help "compile/analyze and plot the outcome" so it take bulk of the heavy lifting.

However, most of the stuff I do use, I already know typically what kind of outcome I am expecting, so if the output is way out of wack I know something is wrong, or the expected outcome has changed over time, which would promote another review at what needs to change.

Overall, good and controlled use of AI can help your work, personally and professionally, but I would just caution to not totally depend on it, as others have said it, the learning curve is still very steep there.

 

I've been playing with a no-code automation tool called N8N. Has an extensive API backlog. Really easy and intuitive. This was the gateway to learning about AI agents after getting RAG going with vector DBs and some chat interfaces.

The next big news is Pydantic-AI. Currently reading its docs and will be trying this as agentic usage outside of language tasks has been annoying.