I’m not sure if you have some sort of challenge in writing under the time pressure of fast QSOs, but I might write logs every other line and copy the info when there is a moment of break and the memory is still fresh. The second time writing the same info without time pressure is probably more organized or better readable. I see your writing at the top left corner, which I think would probably work.
I personally stop economizing the paper space after this exercise. Dense writing (fewer pages) was slightly easier for manually transcribing the log. That is false economy now, when I can save so much post-activation agony.
I also asked ChatGPT about what I should specifically do to improve the recognition performance. (1) more clear margins and line spaces (2) always write summit/park ref and other stuff at the same position of a page, or delimit them with horizontal rule or box in a consistent style. (3) erase better. I can do that.
Don Sudduth KN0VA just did a comparison between Gemini 1.5 and ChatGPT 4o using the same input image, reported in POTA Slack channel. He said ChatGPT 4o was considerably superior.
What an interesting discussion! Despite being a techy sort, making stuff, writing my own software, etc. I have never thought there was much value in ChatGPT for me.
My limited experiences has been that LLMs seem to be very good at writing entirely plausible-sounding material that stands up to expert scrutiny rather poorly. Certainly the AI stuff that Google now offers in response to my carefully crafted search phrases generally leaves much to be desired. I guess I’m just an old fogey that actually likes writing his own software and positively enjoys word-smithing! Maybe I need to get out more…
For logging QSOs when out and about I record the audio on my mobile phone, via a splitter from my headphones socket. Back home I transcribe the audio into my computer log at leisure, listening intently to the interesting bits of QSOs and skipping over the waffle.
That was certainly my experience prior to GPT 4o and o1 models that came out in the fall. I’m slowly changing my attitude toward it, but I had already decided worth paying to remove the access limits a while ago. As I said, asking GPT something about RF engineering or antenna is still hugely disappointing or amusing.
This thing does not come with clear instruction as to what it can do or how it should be used, and unlike Apple products, it’s unclear what to do with it. So the value of ChatGPT depends a lot on how good you are in engineering the prompts.
I have developed a bit of a routine with logging. Once I’m off the hill, I will call into a pub, either on the way home or later that evening. While enjoying a pint, I transcribe my log into Ham2K on my ‘phone.
I have one too. When I get home after an activation I sort out all my gear - hang up jackets, gaiters, tarp etc, dismantle my pole and air my antennas and other kit. Then I make a big mug of tea and while I drink that I sit at the PC and type my log into FLE. Upload to SOTA, POTA, Log4OM. Update my spreadsheet.
I can’t see that GPT would save me any time. By the time I’d checked the log I could have typed it in myself.
Me too.
The post activation ritual of deciphering my handwritten logs, updating the database and exploring who has chased me, and where. A slow time natural rounding off of the day’s adventure; rigours, successes and failures completes the event. Attention to the process of the day, its content and experience are part of my SOTA experience, the gift that keeps on giving. Yes, it can be a chore at times, collating and entering the data but it validates and completes the effort made to start and finish the mission. Sometimes it would be nice if someone did it it for me but, it would, for me be at the loss of a vital aspect and component of the process.
I should try this ‘drying my pole’ business sometime. I had mine frozen solid on a recent activation in mid-Wales…I thought I was going to have to walk back to the van with it - in the dark and fully extended
The trailheads lie a grueling 2 to 3 hours’ drive away, and I make the most of the effort by lingering on summits to fill several pages of my log. In the summer, I often tackle two or three peaks to justify the 4+ hours spent on the road.
One ritual keeps me steady: brewing a large cup of tea or coffee in the trailhead parking lot before the long drive home. The highways, dull and unchanging, require vigilance, though the brief stretch of dirt roads offers a fleeting thrill. By the time I return, exhaustion has overtaken me, and the log waits until morning—accompanied by two strong cups of coffee to push through. It’s the official tax paperwork—mandatory, tedious, and utterly inescapable—a grim requirement to keep playing the game, whether you enjoy it or not.
Another element in the mix is this: AI is not perfect, but it is continually evolving. Its flaws do not diminish its usefulness in my productivity. The key lies in adapting my routine and protocols to extract the maximum utility from its capabilities, flaws and all. For instance, I adjusted my prompts to enclose sections where the OCR stumbles within [brackets], allowing for manual correction later. It even provides a concise, well-organized report at the end, highlighting areas that need attention.
When I bring POTA into the mix, the OCR workflow promises industrial-scale savings.
My current prompt:
Please type out all the info you pick up from this amateur radio log in FLE format. FLE format does not use comma to delimit columns, just spaces or tabs. It also uses hhmm format not hh:mm. Frequencies are in the format of n.nnnnn where the number of digits vary. Please do the entire pages, do not ignore or omit parts you are uncertain. Indicate where you are not certain in brackets, and bring them to my attention, but never ignore or omit. Enclose the QTH and other remarks after the signal reports in <> brackets. Copy the result in text buffer.
Most of us XY types have a ready supply of and easily targetted source of warm liquid with us that can be used in such emergencies to melt frozen poles.
I agree about the benefits of the post activation events including the manual transfer of my [completely un-AI-readable] handwritten log into FLE and consideration of who chased me this time.
Oh definitely. If my 6m pole gets the slightest bit damp [which is most of the time in NW England] I leave it fully extended overnight in a warm room - ditto the rest of the radio kit (taken out of their drybags).
If I have trouble collapsing any of the pole sections on summit, I disassemble the entire pole, wash the sections in soapy water, rinse and dry, and reassemble the pole the next day.
