This is a guest post by Rich, 4X1DA, exploring how AI in Ham Radio is revolutionizing the hobby.
Imagine having your own personal RF/antenna engineer, IT expert, Elmer, and all-around technical genius—on call 24/7.
Not the kind that just spits out facts like a glorified Google Search, but someone you can have a real conversation with. Someone who listens, understands, and responds thoughtfully—offering not just answers, but ideas, improvements, and insights you hadn’t even considered.
She speaks your language—literally. Want a Midwestern twang? Done. Prefer an Indian accent? No problem. You don’t even have to type—just talk. And the more you interact with her, the smarter she gets.
Meet your new genie: ChatGPT.
She’s not magic—she’s a Large Language Model (LLM), a powerful form of machine learning (ML), which itself falls under the broader umbrella of artificial intelligence (AI). And while she’s not the only digital brain in town, she’s currently the most popular, holding over 60% of the LLM market share just two years after launch.
The graphic below shows the most popular LLM contenders in mid-2025.
Of course ML, ChatGPT or other LLMs weren’t specifically designed for Ham Radio but these tools can help unlock new levels of education, efficiency, experimentation, and accessibility in our hobby.
The secret of using AI is knowing how to ask the right questions and being able to separate the signal from the noise.
Below are six practical examples of how AI is being used with Ham Radio – today:
Antenna Design Assistance from ChatGPT
Instead of flipping through the ARRL Antenna Book, many Hams now ask ChatGPT for help with designing antennas. The key is to provide detailed input to avoid the ‘garbage in, garbage out’ problem. ChatGPT can suggest design parameters and even offer follow-up questions, like “Would you like me to display the radiation pattern of this antenna?” You must still be careful when using AI. It can “hallucinate,” meaning it can confidently provide incorrect answers. Always cross-check its suggestions with other large language models (LLMs) like Gemini, DeepSeek, Claude, MS Copilot, and Grok.
I recently thought about raising my VHF omni antenna height from 3 meters to 6 meters above the roof. Is it worth the effort and expense? Find out the surprising answer I discovered from ChatGPT here. TLDR: No, it’s not worth it.
Another potential project: Replace my 40m Inverted V with a Delta Loop or a Groundplane antenna. Again, ChatGPT is perfect for this kind of comparison and deeper analysis of what these antennas are capable of. More on that here.
The interactive dialogue with ChatGPT makes antenna design fun while encouraging experimentation and deeper learning.
Transcribing and Searching Audio with AI
One of the most practical AI applications for Hams today is audio transcription. Tools like Turboscribe (free, for 3 uploads a day) can automatically transcribe recorded audio files, turning long nets or contest recordings into searchable text. For example, a Ham could upload a recording of a weekly net to Turboscribe, then use a simple script to search the transcript for a ‘keyword’ like “DX.” This capability saves operators hours of manual listening by instantly pinpointing segments of interest. Here is an example of a recorded 2 meter Simplex QSO between two local Hams.
But the magic doesn’t stop there. The transcribed text can be uploaded to Google’s NotebookLM, an AI program that summarizes and simplifies complex documents, videos, web pages or even entire books. Recently, NotebookLM added a feature to convert these summaries into educational-style videos, giving Hams a fresh way to digest dense technical material or recaps of lengthy on-air discussions. Here’s an example of how NotebookLM summarized a YouTube video of “An Introduction to Ham Radio” by 4Z1ZV, into a user-friendly podcast format (in Hebrew).
Noise Reduction and Signal Enhancement via AI Servers
Another breakthrough is AI-powered noise reduction/filtering. Instead of relying solely on local DSP filters in transceivers, applications like RM Noise stream audio to remote AI servers that process the signal in real time. For SSB, this can produce near-FM quality audio, dramatically reducing static and interference. For CW, AI can virtually eliminate background noise on even very weak signals, making the incoming code sound as clean as if it were generated by a perfect code oscillator. This technology is a game changer for operators struggling with challenging band conditions or urban noise. Learn more about RM Noise here.
Data Analysis of Digital Modes with AI
Digital modes like WSPR and FT8 generate vast amounts of data that can be overwhelming to analyze manually. A more clever approach is to save decoded data to an Excel file and then upload the spreadsheet to AI platforms like ChatGPT or DeepSeek. Within seconds, these tools can analyze the data and provide detailed insights such as: furthest and closest stations heard, most commonly logged countries, rare DX spots, propagation trends by time of day, plus much more. This rapid, AI-driven data crunching helps operators better understand band openings and optimize their operating schedules. Here’s a summary analysis of the WSPR 20M data for 12 July 2025
Circuit Design and Reading Schematics
ChatGPT is capable of creating or reading schematic diagrams. I recently used it to suggest improvements for an old ICOM IC-735. Here’s a snippet from the dialog:
Sure! Please upload the schematic of your ICOM IC-735, and I’ll analyze it for improvements such as:
- Signal clarity & labeling (e.g., adding missing component values, clearer net labeling).
- Modern component replacements (substitutes for obsolete parts, better efficiency or noise performance).
- Power supply & grounding enhancements (reduce noise or hum).
- Layout optimizations (reducing coupling, improving shielding).
- Service documentation improvements (annotations, test points).
Once you upload it, I can highlight weak points, redraw sections more clearly, and recommend specific upgrades (like swapping op-amps, RF transistors, or caps for modern equivalents).
Can you upload the schematic?
Robo-QSOs: AI as an Active Operator
A fascinating frontier is AI-driven automated QSOs. Recently, a Ham in Belgium, ON7OFF, set up a system in which AI acts as a “robo-QSO” operator. The AI-generated voice calls “CQ-DX” with the operator’s callsign, listens for replies, converts received audio to text, and carries out a conversation in perfect, non-accented English. It even reads the S-meter to provide accurate signal reports. After the exchange, it logs the QSO and uploads the contact information directly to popular logging software like Log4OM. Watching this seamless interaction between AI and human operators feels like witnessing the future of Ham Radio in real time. I haven’t tried this myself as I’m concerned about the ethics of Robo-QSOs and the very real possibility of this type of technology being used by bad actors in contests or chasing awards like DXCC.
Summary: AI – A New Era, also for Ham Radio
This article only scratches the surface of what is possible. There’s so much more to explore and discover!
If you haven’t done so already, head over to OpenAI/ChatGPT and start a conversation. The first thing you will notice is that ‘the machine’ has endless knowledge but never an opinion (how refreshing 🙂).
Be sure to compare feedback with other AI platforms mentioned in the beginning of this article since AI can ‘hallucinate’ and come up with ‘facts’ that range from outright wrong to hilariously funny.
From transcribing net recordings and cleaning up noisy signals, to analyzing WSPR and FT8 data and operating robo-QSOs, AI and machine learning are revolutionizing how Hams communicate, learn, and experiment. ChatGPT and other AI tools are empowering operators to transcend traditional boundaries, making the hobby more efficient, accessible, and exciting than ever.
Click here to listen to an AI generated audio summary of this entire article (~14 minutes)
