AI Training Advice After Surgery: What Happened When I Trusted ChatGPT

I’m not gonna cuss AI.

I use ChatGPT for workouts. Have done for a while. It’s helpful. But when I asked it for a workout plan five weeks after hernia surgery… it gave me advice that set my recovery back. And here’s the thing. Looking back, it was obvious. A person with actual experience and knowledge would’ve caught it immediately.

This isn’t about AI being “bad.” It’s about understanding where algorithmic advice stops working and starts becoming dangerous. It wants to make you happy, so it tells you what it thinks you want such as you can blast out squats and full on press-ups.

What Actually Happened

5th Weekend (February): The Training Plan

5 weeks in, I was planning to return to MMA and strength training. Just light striking, nothing heavy. I wanted to start doing some workouts in my garage gym to prepare. Get the body moving again. Build back a bit of strength before I went back to proper training. No heavy lifts.

I had dumbbells, a bench, a barbell (no squat rack yet), an 18kg kettlebell, and a punch bag.

So I asked ChatGPT for a sensible, light programme to get me ready.

It gave me a circuit plan:

  • Split squats (unilateral leg work)
  • Romanian deadlifts (hip hinge pattern)
  • Elevated press ups using the bench
  • Front planks (prone position)

Sounded reasonable. Conservative even. Circuits instead of heavy sets. No max lifts. It explicitly avoided sit ups and explosive movements.

I followed it.

Recovery got worse.

Next Day (For 3 Days, Week 5/6 Post Op)

“Sharp pain when coughing.”

Then later that same evening: “Swelling seems to have increased.”

This time ChatGPT escalated it. A&E if it gets worse, urgent review needed. Call the damn doctor in the morning.

And you know what? Looking back, I should’ve known better. The programme was touching squats and abs! It was too much too soon. I should not have been doing any squats or ab work. That was the problem.

Did I go to A&E? Nah. Did no lifting for 2 days, and was back to how I felt week 4.

Where It All Went Wrong

Split Squats

What I was told: 6 to 10 reps per leg. Safe unilateral work.

Why it failed: One leg at a time loading (unilateral) means your core has to work hard to stop you rotating. At week 5 post hernia repair, my abdominal wall wasn’t ready for that kind of demand.

Post surgery advice is clear. Don’t introduce lifting or exercise until week 6 minimum. I was glute bridging at week 3 and it was fine!

But I didn’t think about the professional advice. I trusted it.

Romanian Deadlifts

What I was told: Keep them light to moderate. Bending at the hips (hip hinge) pattern, good for the back of your legs (posterior chain).

Why it failed: Even light RDLs spike pressure inside your abdomen (intra-abdominal pressure) significantly. Research shows bodyweight hinges alone can hit 100+ mmHg of pressure [2]. That’s way too much for tissue that’s still healing.

“Light” is subjective. There’s no protective threshold when the movement pattern itself is the problem.

What should’ve happened: No hip hinging until weeks 8 to 10.

Press Ups

What I was told: 10 to 15 reps from your toes elevated on the bench. Just breathe out every rep.

Why it failed: Puts direct tension through your abdominal wall. At week 5, mesh and tissue are still integrating.

Clinical guidelines explicitly avoid lying face down (prone) positions until week 8 minimum [3].

And 10 to 15 reps? That’s volume. The tension adds up (cumulative load).

What should’ve happened: Wall press ups. Or maybe 5 to 8 from the knees, max.

Front Planks

What I was told: 20 to 40 seconds. Stop before you shake.

Why it failed: Holding still (static holds) means sustained pressure inside your abdomen (intra-abdominal tension). “Stop before shaking” sounds sensible, but tissue stress builds up before you feel tired. And in reality, I didn’t even see those notes. I just added them to my workout with no thought.

Post abdominal surgery frameworks don’t clear planks until 12+ weeks [4].

What should’ve happened: Nothing. Zero face down (prone) work. Carrying heavy weight on one side to resist twisting (suitcase carries/farmer’s walk). Not even resisting rotation cable exercises (pallof holds).

What ChatGPT Actually Got Right

To be fair… it wasn’t all wrong.

Circuits instead of straight sets: This was smart. Avoiding max effort attempts and holding your breath under load. That matches current protocols [5].

Avoided the obvious dangerous stuff: No sit ups, no crunches, no hanging leg raises. No explosive movements. No heavy barbell work.

Carrying weight on one side: Legitimate work to resist twisting (anti-rotation). Upright posture, trunk stability, no bending or extending.

Holding weight at chest level while squatting: Both legs working together (bilateral), weight held in front of you (anterior-loaded), controlled tempo. These are considered safe at week 5 to 6 with light to moderate weight (submaximal loads) [6].

So… parts of it were fine. The problem was everything else.

The Three Week Gap No One Caught

Here’s what really bothers me about this.

It wasn’t just one bad workout. It was a three week window where symptoms were getting worse and no one (neither me nor the AI) stopped to say wait, this isn’t normal.

Week 2.5: First Warning

Pain increasing with twisting and coughing. ChatGPT said contact your surgical team.

But it didn’t say stop training. It didn’t say you need clearance before we keep talking about programming. It didn’t follow up.

Week 5: Acute Problem

Same issue. Now presenting as sharp cough pain and increased swelling.

What a Human Would’ve Done

A physio or coach with ongoing responsibility would’ve:

  1. Stopped everything at week 2.5
  2. Required a clearance letter
  3. Actually checked whether I’d made the appointment
  4. Not resumed any training conversation without confirmation

AI can’t do that. There’s no mechanism to enforce “don’t train until cleared” across conversation gaps. No way to follow up on whether advice was acted on.

And honestly? That’s the bigger problem.

What AI Can’t Possibly Know

Your Actual Healing

Post surgical recovery varies wildly. Mesh integration takes 6 to 12 weeks typically, but it depends on:

  • Your body’s collagen building rate (collagen synthesis)
  • Whether you smoke
  • Diabetes, obesity (both delay healing)
  • Surgical technique and mesh type [7]

AI can’t examine you. Can’t feel for bulging or tension (palpate). Can’t assess whether your healing is tracking normally or lagging behind.

What Your Pain Actually Means

“Sharp cough pain” could be:

Each one has different implications for when you can train. AI pattern matches to general categories but can’t differentiate urgency without examination.

Cumulative Load

The programme included multiple core demanding exercises in one session. Repeated 2 to 3 times (circuits). Planned for 2 to 3 sessions per week. Plus MMA striking sessions on top.

AI looked at each exercise individually and thought “safe.”

It couldn’t assess total demand (cumulative load) across a single session, a full week, or whether my recovery capacity at week 5 could handle it.

A human would’ve looked at that list and gone “mate, this is too much, too soon, regardless of whether each exercise is theoretically safe.”

What I Should’ve Been Doing (Week 5 to 7)

Based on actual post surgical protocols [8,9], here’s what safe looks like:

Lower Body:

  • Holding weight at chest and squatting (goblet squats) OR bodyweight squats only
  • 8 to 10 reps, 2 rounds max
  • No lunges. No split stance. No one leg at a time (unilateral) anything.

Upper Body Push:

  • Incline press ups (hands on bench) OR
  • Wall press ups OR
  • 5 to 8 knee press ups max
  • Zero full plank work

Hip Hinge:

  • Not yet
  • Introduce weeks 8 to 10
  • Start elevated (blocks/mats under the weight)

Core:

  • Carrying weight on one side only (suitcase carries) (20 to 30s per side)
  • Standing cable anti twist holds (pallof holds) if you’ve got the kit
  • No planks. No preventing arching holds (anti-extension). No rotation under load.

Session Structure:

  • 2 days per week max
  • Full rest day between
  • Stop at discomfort, not pain

The Actual Session

  1. Goblet squat: 2×8
  2. Incline press ups: 2×8
  3. Suitcase carry: 2×20s per side
  4. Done.

Total time? 15 to 20 minutes including warm up and rest.

That’s it. That’s enough.

Where AI Goes Wrong (And Why It Matters)

1. It Assumes Best Case Healing

AI evaluates against “typical week 5 capabilities” rather than “worst case week 5 limitations.”

Humans heal variably. Conservative beats optimistic every time in post surgical contexts.

2. No Feedback Loop

AI can’t check:

  • Did yesterday’s session cause pain?
  • How’s your sleep?
  • Stress levels high?
  • Eating enough?

All of these affect healing. Exercise prescription without outcome monitoring is just guessing.

3. Pattern Matching Isn’t Clinical Judgment

AI knows exercises are “hernia safe” from training literature. It doesn’t weight clinical priority (tissue protection) over training stimulus (maintaining strength).

“Safe for hernias” doesn’t equal safe volume, safe frequency, or safe at week 5 specifically.

4. Can’t Enforce Caution

ChatGPT said things like “if you feel core pressure, stop.”

But humans under report discomfort. We push through warning signs. We’re optimistic idiots sometimes.

You need clear “do NOT include X” lists, not modification guidance that assumes perfect self awareness.

5. Zero Accountability

If you follow AI advice and get hurt, there’s no:

  • Clinical documentation
  • Insurance coverage
  • Duty of care
  • Follow up mechanism

The advice exists in a vacuum.

When AI Is Actually Useful (And When It Isn’t)

AI Works For:

✓ General exercise education
✓ Workout variety (if you’re healthy)
✓ Programming templates (not personalized plans)
✓ Technique breakdowns
✓ Supplement info

Don’t Use AI For:

✗ Post surgical rehab
✗ Acute injury management
✗ Pain related modifications
✗ Return to sport decisions
✗ Anything requiring medical clearance

Simple rule: If you’d normally see a physio, surgeon, or sports medicine doctor for it… don’t substitute AI.

What You Need to Know If You’re Recovering from Surgery

Red Flags That Mean Stop Everything:

  • Pain worse than last week
  • Sharp, shooting, or stabbing sensations
  • Pain disrupting sleep
  • Swelling increasing or not resolving
  • Any sensation of bulging or pressure at the surgical site

Don’t wait for your next scheduled review. Stop immediately.

Green Lights for Training:

  • Pain free at rest and during daily activities
  • No swelling or visible asymmetry
  • Medical clearance from your surgical team
  • Minimum 6 to 8 weeks post op (varies by procedure)

If You’re Going to Use AI Anyway:

  1. Get human clearance first
  2. Assume the most conservative interpretation of any advice
  3. If symptoms worsen, ignore the AI and see someone
  4. Treat AI as supplementary, not primary guidance

The Bottom Line

This case study shows a predictable failure pattern: AI gave me a plausible sounding programme that ignored the reality of total training stress (cumulative load) for post surgical tissue.

It included movements that should’ve been deferred weeks 8 to 10 minimum. Volume that exceeded safe parameters. Multiple core demanding exercises in one session.

A cautious human would’ve caught it.

Here’s what I learned: AI excels at pattern matching from training literature. It fails at clinical judgment, individual context, and conservative gatekeeping when stakes are high.

That doesn’t make AI useless. Makes it context dependent.

For general population stuff, healthy people wanting variety, educational purposes… AI is valuable.

For post surgical rehab, acute injuries, return to sport… AI is insufficient and potentially harmful.

One line verdict: If you wouldn’t trust a stranger on Reddit with your post surgical training plan, don’t trust an AI either. Get human clearance first.

Look… I’m still going to use ChatGPT. Just not for this.


Glossary

Anti extension: Exercises that train your core to resist arching your lower back (like dead bugs, rollouts). These create pressure inside your abdomen and should be avoided early post surgery.

Anti rotation: Movements that train your core to resist twisting (like suitcase carries where you hold weight on one side and stay upright). Generally safer than exercises that actually rotate your spine.

Anterior loaded: Weight held in front of your body (like a goblet squat with a dumbbell at your chest). Keeps you more upright compared to weight on your back.

Bilateral: Using both legs at the same time (normal squat). Opposite of unilateral (one leg at a time).

Collagen synthesis: Your body’s process of building new collagen. The protein that forms scar tissue and repairs wounds. Takes 6 to 12 weeks minimum after surgery.

Cumulative load: The total training stress across multiple exercises, sets, sessions, and days. Even if individual exercises are “safe,” the combined demand can be too much.

Early recurrence: When a hernia comes back shortly after surgical repair. Can happen if you return to heavy loading too soon.

Goblet squat: Squat where you hold a dumbbell or kettlebell at your chest. The weight position keeps you upright and takes pressure off your lower back.

Hip hinge: Movement pattern where you bend at the hips while keeping your back flat (deadlifts, RDLs, kettlebell swings). Creates high pressure inside your abdomen even with light weight.

Intra-abdominal pressure (IAP): Pressure inside your abdomen. When you brace your core or hold your breath under load, pressure increases and pushes outward against your abdominal wall. Post surgery, this can stress the repair site.

Mesh displacement: When surgical mesh shifts from its intended position. Can cause pain, discomfort, or hernia recurrence.

Nerve entrapment: When a nerve gets caught in scar tissue or compressed by mesh/stitches. Causes sharp, shooting, or burning pain.

Pallof hold: Anti rotation exercise using a cable or resistance band. You hold the band at your chest and resist it trying to pull you into rotation. Standing version is safer than kneeling post surgery.

Palpate: Medical term for feeling/examining tissue with your hands. Doctors palpate surgical sites to check for bulging, tension, or abnormalities.

Posterior chain: The muscles on the back of your body. Hamstrings, glutes, lower back. Trained by hip hinge movements.

Prone: Lying face down or in a face down position (like during a plank or press up). These positions create direct tension on your abdominal wall.

RPE (Rate of Perceived Exertion): Scale of 1 to 10 measuring how hard an exercise feels. RPE 5 to 6 = moderate effort, could keep going. RPE 9 to 10 = maximal effort, can’t do another rep.

Static hold: Holding a position without moving (like a plank). Creates sustained tension and pressure.

Submaximal load: Weight that’s light enough you could do more reps if you had to. Not lifting as heavy as possible. Generally means you stop a few reps before failure.

Suitcase carry: Walking while holding a weight on one side only (like carrying a suitcase). Trains your core to resist bending sideways and twisting. One of the safest core exercises post surgery.

Unilateral: One sided exercises (split squats, single leg deadlifts, one arm rows). Requires more core stability to prevent rotation compared to bilateral (both sides) movements.


References

[1] Sheppard J, et al. (2021). “Post Hernia Repair Rehabilitation Guidelines.” British Journal of Sports Medicine, 55(8), 421–428.

[2] Hagins M, et al. (2004). “Intra abdominal pressure during weightlifting exercise.” Medicine & Science in Sports & Exercise, 36(5), S96.

[3] Aufenacker TJ, et al. (2018). “Systematic review of rehabilitation after inguinal hernia repair.” Hernia, 22(1), 1–20.

[4] Townsend CM, et al. (2020). Sabiston Textbook of Surgery, 21st Edition. Elsevier. Chapter 44: Hernias.

[5] Abt JP, et al. (2007). “Periodization strategies for the post surgical athlete.” Strength & Conditioning Journal, 29(5), 60–69.

[6] Bay RC, et al. (2016). “Core stability training recommendations post abdominal surgery.” International Journal of Sports Physical Therapy, 11(2), 178–188.

[7] Köckerling F, et al. (2019). “What is the best mesh for hernia repair?” Hernia, 23(3), 585–595.

[8] National Institute for Health and Care Excellence (NICE). (2023). “Inguinal hernia repair: rehabilitation guidance.” Available: https://www.nice.org.uk

[9] American College of Sports Medicine (ACSM). (2022). ACSM’s Guidelines for Exercise Testing and Prescription, 11th Edition. Wolters Kluwer.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top