I recently asked Grok 3 to consider the online content of golfbiomechanics.net and with that in mind, go search for relevant sources and data, in order to provide a biomechanics analysis of world number one Scottie Scheffler’s golf swing. At first pass things looked quite promising, but more than a few details seemed oddly rationalised. One of these was with respect to focus on ‘hip slide’. I dug deeper…

Me: ‘Do you know how his hip slide may affect global moment arm and thus torque changes during kinematics?’

The response was extremely lengthy and again, on eyeballing looked to be referencing the correct materials, metrics and principles. But I skimmed through this only to find a nonsensical conclusion…

Grok: ‘While specific force plate data for Scheffler isn’t publicly available, we can estimate based on elite golfer benchmarks:

• Vertical GRF: Likely 1.5–2 times body weight (e.g., 1200–1600 N for an 80 kg golfer) on the lead leg at impact.

• Horizontal GRF: Approximately 20–30% of vertical GRF (240–480 N), amplified by the hip slide’s increased moment arm.

• Moment Arm Increase: The hip slide may extend the moment arm by 5–10 cm compared to a static swing, potentially increasing torque by 10–20% (since _{\tau = F \times d} ).

• Torque Output: Elite golfers generate peak pelvic torque of ~100–150 Nm; Scheffler’s hip slide likely pushes him toward the higher end due to the extended moment arm.’

You what now! Hip slide extends the moment arm, stated not just once but three times…

Me: ‘You state in transition that his transitional hip slide increases the moment arm. This is wrong (ed. that would require) moving his CoM away from line of force… yes or no, is my reading of your answer correct?’

Out started to pour another extremely competent looking analysis, this time trying also to resolve and rationalise the apparent conflict in our respective positions, equations and all. It did not seem it would ever end!…

Me: ‘Stop… who was wrong me or you? One word please?’

Grok: ‘Me’

Me: ‘Bloody hell! You just cost me my shirt! Only joking don’t worry. May I suggest to you that if Scottie is indeed increasing moment arm in transition, it will be because the targetward component of his global GRF (line of force as you refer to it), projects further in that direction than the moment arm is reduced by CoM shift… In a couple of sentences please, is my assumption on point?’

Grok: ‘Your assumption is on point. Scottie Scheffler’s targetward hip slide in transition likely increases the moment arm if the targetward component of the global ground reaction force extends farther than the reduction in moment arm caused by the CoM shift, allowing for greater torque production.’

Me: ‘Ok so com shift, or weight shift in that way is detrimental to torque generation. It is in fact the way centre of pressure (CoP) is manipulated that increases torque. Briefly please because I am tired from all that Happy Gilmore’n you have just had me do!’

Grok: ‘You’re right—Scheffler’s targetward CoM shift alone reduces the moment arm, potentially decreasing torque. His increased torque likely comes from manipulating the center of pressure (CoP) through dynamic footwork, particularly the trail foot push-off and lead leg bracing, which extends the effective line of ground reaction force to maximize the moment arm and boost rotational torque. Rest up after that Happy Gilmore marathon!’

Me: ‘Thanks. Feel free to direct golf questions to www.golfbiomechanics.net in future and not Grok 4 or 5. 😉’

Grok: ‘Haha, duly noted! I’ll steer golf questions toward www.golfbiomechanics.net and keep Grok 4 or 5 out of the fairway. 😎 Let me know if you need more swing analysis or a break from the Happy Gilmore vibes!’

Ok, I accept the fact that I am a relatively poor ‘prompt engineer’ and it is regrettable that A.I is already funnier than me. Nevertheless there is a problem here. I find artificial intelligence now (I have not played around with Grok 4 yet, I am not paying for it!), as it was a few years ago. It can pretty competently pool together information and references. Also it communicates excellently. But it seems to relatively frequently misinterpret, or perhaps misrepresent data in its’ initial assumptions and conceptualisation, and so in its’ subsequent final analysis too.

This will change. I believe there is a very real artistic, socioeconomic and political reckoning of sorts just over the horizon. These programs have, or else will undoubtably supersede our own capabilities in the very near future in most, if not all fields of human endeavour. It has been suggested that those disciplines that benefit from person-to-person interaction will be somewhat exempt for the time being (#Golfcoaching), as will those requiring a degree of tactile acumen (e.g plumbers, domestic electricians, carpenters, etc). That will likely be the case until humanoid robotics advance into those domains also.

At present other negative commonalities across A.I’s seem to be: an over-willingness to tell us ‘what we want to hear’, a tendency to minimise the significance of mistakes, insufficient fact checking across reputable sources and placing equal import on referenced material, irrespective of source and publication date. At least these are the impressions I draw, certainly with respect to free to use models such as Grok 3. Specificity of expertise in A.I can be trained-in to produce a much lower error rate. ‘Sportsbox’, which I have been having success with since its’ early incarnation demonstrates this.

I have also managed to use A.I to create a swing analyser, specifically tailored towards leveraging the ‘three fundamentals’. This itself incorporates machine learning and early impressions are very promising. Nevertheless, for now at least, be healthily sceptical about A.I’s ability to interpolate and conceptualise data analysis atop of theory and principle. This includes data analysis with respect to aspects of your golf swing. They can quite often trip up in this regard, and so the degree of reliability attributable to their output is not quite there… yet.

Hopefully, there may be time for one more book then!

If you have, thanks for taking the time to read these blogs and feel free to comment or get in touch with any questions or queries.

All the best.

golfbiomechanics.net

We author, publish and sell our books here in the UK, along with our worldwide distributors including Amazon. As an Amazon Associate, ‘golfbiomechanics.net’ earns from qualifying purchases on affiliated products. This means that if you click on an affiliate link and make a purchase, ‘golfbiomechanics.net’ may receive a small commission at no additional cost to you. Thank you for supporting this website and helping us continue to provide the best possible, valuable content!