Beckett probably heard those very words from his wife. Interestingly enough, ChatGPT wasn't quite sure about the question either, but that didn't hinder it from offering quite an elaborate reply. It also has a detailed response to "What will happen when 85% of all human beings are obsolete and unemployable?" and specifically mentions AI as being a possible cause. Personally, I was picturing some combination of when Burning Man got rained out and current Haiti.
AI is fantastic for giving me enough information about a random topic to begin asking intelligent questions during a QSO. I don't care about accuracy of details; that's for the QSO to handle.
Congrats on being one of the few who actually TALKS during a QSO-- a relatively rare occurance except for aome polluted band portions on 40 and 80 m....
Will toss in my quick 2c. We use LLM's quite often in our work, but the key as previously pointed out here is training data / data sets and narrowing the application of the LLM to I/O acceleration -- driving efficiency in routine elements of accessing and coordinating data and workflows and helping optimize them (vs having it BS about the BS found on the internet). It's quite incredible in this regard as a productivity tool. For learning and engaging with topics in ham radio, one of the ways I've enjoyed interfacing w AI is using Google's Notebook LLM. Specifically, if you have sets of PDF documents such as extensive manuals, learning guides or detailed handbooks, you can upload a set of these as the reference documents and it will interact with you strictly based on the content of those materials. This has been fairly useful in accelerating learning, spec'ing out antenna projects, etc. For fun, you can even click a button and have it turn some/all of the materials into podcasts (not practical, but a fascinating experience the first time you experience it). For coding, Claude has been a fun tool to build some interesting tools very quickly that I wouldn't even have attempted if I had to code from scratch. YMMV but fun to experiment with.
Heres a use case for chatgpt that ive used to learn Arduino coding. If you ask chatgpt "write an arduino program to energize an output pin when an input pin goes low, or maybe when an analog pin is above 4 volts.". It will generate the code. Ask it additional question based on your desired program use and you can get additional code examples. After a while these examples you can begin to understand how to program one of these. I did something similar to make a screwdriver antenna controller with some neat capabilities and it really helped. It didnt write the entire program, but i took pieces of code to ultimately build a program and i made the tweaks to make it work the way i wanted. Bottom line, some of this AI stuff you can use to learn something new. Though I still worry about what chatgpt will end up doing to future jobs and how much future generations will actually learn something when term papers could probably be written in minutes with chatgpt
This is the greatest thing since sliced bread! At least for using SimNEC. I hate trying to use SimNEC and I know that I am not alone. To me, it is more of a programming environment than an actual application. Programmers like it. I don't. I LOVE what it does, it is invaluable! And my hat is off to Ward for creating and growing it. But I hate trying to use it. One of the reasons I find it frustrating is the way in which it is documented. Searches in the manual for obvious things often turn up nothing or tangential minutia that do not explain the topic. It has been highly frustrating. HOWEVER, I loaded the SimNEC manual and the ANVIL programming guide into Google NotebookLM and now I ask it questions. It gives me usable answers in short order. It pulls together relevant info no matter where it gets mentioned in the manual. I no longer dread trying to do something with SimNEC. What a breath of fresh air! This issue is not helped by the fact that I hate hate hate sitting through interminable videos that purport to explain how to use software. They are most often out of date, slow as hell, and usually don't give me what I need but give me a bunch of crap I'm not interested in (yet) or make unexplained references that assume the viewer has watched every previous video that has ever been made on that topic. This is not just an issue with SimSmith videos. IMHO, YT video is a lousy medium for teaching people to use software. I really cannot make myself sit through many of them. So being able to query the documentation really helps me. I think AI digestion- summation-query IS the proper medium to teach people to use software. Examples that are fully notated and explained would help too. NotebookLM facilitates that. I cannot thank you enough for turning me on to NotebookLM! Thanks, Mike WY6K
Normaly its worth in a pdf manual to use Ctrl + F. Unfortunately its not always possible. When old manuals are just scanned and the PDF only contains pictures. Here AI got to find and choose another source. And when its really fast to cruise the Internet well, AI is a big help.
The Find function is lame because it turns up every mention of the term and then you have to sort through them to find what you want. For instance, the search function in some software manual does not work with muliple words, i.e. "syntax for the Plot function". Instead it either returns null, or a list of the 132 times the word Plot is used as well as the 141 times the word function is used. Search or "Find" functions just need to be smarter and AI like this is the key. Photos and handwritten material, like "pencilled out" equations - much less cursive writing - no doubt remain challenges. But given enough time and money, those will probably get sorted out too.
HERE YOU GO: Summary of the YouTube Video: “Using Artificial Intelligence in Your Ham Shack” This video explores how AI, specifically large language models (LLMs) like ChatGPT, can be utilized in amateur radio (ham shack) environments. The presenter demonstrates various applications and benefits of AI for ham radio enthusiasts. Key Highlights: • Introduction to AI in Ham Shack: • AI in this context refers to LLMs that can summarize text, generate content, and answer questions. • The focus is on simplifying tasks and enhancing efficiency. • Examples of AI Applications: 1. Building Antennas: • Demonstrated how ChatGPT provides step-by-step instructions for creating a 40-meter dipole antenna. • AI offers material lists, formulas, and tips for safety and construction. 2. Advanced Antenna Projects: • Detailed guidance on building a 40 to 10-meter end-fed half-wave antenna. • Included troubleshooting advice for issues like high SWR, with suggestions for checking components like transformers, antenna length, grounding, and deployment. • Schematic Analysis: • AI can interpret schematics, such as those of old tube radios, and provide troubleshooting steps. • For example, it identified issues related to a coupling capacitor and offered solutions. • Image Recognition and Guidance: • The AI can analyze images of radios or schematics and answer related questions. • Demonstrated with both vintage and modern radios, providing operational and tuning instructions. • For instance, AI explained how to tune a transmitter and set a notch filter in modern radios. • Benefits of AI in Ham Radio: • AI serves as a virtual mentor (“Elmer”) available 24/7. • It streamlines tasks like building, troubleshooting, and operating radios without needing extensive manual searches. Conclusion: AI tools like ChatGPT have immense potential in ham radio setups, from learning and building to troubleshooting and operating equipment. Viewers are encouraged to share how they would use AI in their ham shacks.
I’ve used chatgpt a few times when making images for SSTV: “How do I ((insert SSTV-related image making task here)) in GIMP?” It is SO much quicker and easier than goign to the help menu and poring through a user guide, especially if you’re asking it to do something that isn’t explicity laid out in the guide. In my experience, in general, I’ve seen people go from manuals and online user guides to help ticket systems to Youtube videos for getting help on how to do stuff at work and especially with ham radio. Heck, we had one software suite that bypassed all of that and just had a chat channel for reporting problems and seeking help. Now there are many chat channels for tools. It’s just evolution. I predict that the ubiquitous “Help” pulldown menu will eventually just redirect you to chatgpt or another AI capability.
