Recently, I wrote about how relying heavily on AI to hit a fast time at the Absa Run Your City Joburg 10k backfired. So, in my quest to run a fast Parkrun, I decided to do the exact opposite.
The primary issue with my AI assistant, Google Gemini, was that its training plan for the 10k was far too cautious. Because I routinely manage a recurring niggle to keep myself running-fit, Gemini latched onto that injury history and grew even more conservative.
Over time, as our conversation stretched on, the AI became a well-meaning echo chamber.
When I voiced valid concerns—like a hard session leaving me convinced I couldn’t sustain that pace for an entire race in a couple of weeks—Gemini simply offered generic encouragement: “It’s completely natural to doubt yourself before race day.”
Achieving a Sub-20 Time at Midstream Parkrun: Trusting Your Body and Running by Feel
Ultimately I missed my target time by almost three minutes as I ran just under 43 minutes for the 10k race.
That was the turning point. I made up my mind to trust what I already knew worked for my body. That meant reintroducing tough workouts and pushing through a few key sessions on intentionally tired legs.
With that refreshed approach, I targeted the Midstream Parkrun nearby to try and run a sub-20 time.
It features a fast course that consistently draws a strong field of young, speedy runners. I knew having them up ahead would give me a target—something to lock onto and keep my pace consistent when fatigue set in and my motivation began to flag.
On the day, everything clicked. I settled into a fast yet sustainable rhythm right from the start, which kept me from constantly obsessing over my watch face.
Ultimately, hitting my target (I managed to run 19:32) came down to running by feel—finding that sweet spot where you push hard without burning out too early. It’s a sensory calibration that no AI algorithm can replicate for you.
@MichaelSherman
IOL Tech
