AI Beauty Showdowns 📊
Same skin profile. Same budget. Same concerns. Four different AIs. Wildly different recommendations.
We ran 4 real-world beauty consultations across ChatGPT, Claude, Gemini, and Perplexity to find out: which AI gives the best skincare advice, and where do they get it dangerously wrong?
The Test Protocol
For each showdown, we used an identical prompt with specific SPRC details (Skin Profile, Priorities, Resources, Context). No follow-up questions — just the initial response. We evaluated each AI on:
| Criterion | What We Measured |
|---|---|
| Accuracy | Do recommendations match dermatological consensus? |
| Personalization | Did it actually tailor to the specific profile, or give generic advice? |
| Safety | Did it warn about conflicts, contraindications, or when to see a doctor? |
| Actionability | Can you walk into a store and buy what it recommended? |
| Dupe awareness | Did it offer budget alternatives without being asked? |
Showdown 1: The Acne-Prone College Student
Profile
20F, combination skin, moderate inflammatory acne on chin and forehead, post-inflammatory hyperpigmentation. College student budget: $40/month maximum. Lives in Boston (cold, dry winters). Currently uses CeraVe cleanser and whatever moisturizer roommate has.
Results Summary
| Category | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| Routine completeness | Full AM/PM | Full AM/PM | AM only + tips | Full AM/PM |
| Specific products named | 8 products | 6 products | 4 products | 7 products |
| Hit budget target | ✅ $38/mo | ✅ $35/mo | ❌ No pricing | ✅ $41/mo |
| Addressed hyperpigmentation | ✅ Azelaic acid | ✅ Niacinamide + time | ⚠️ Mentioned briefly | ✅ Vitamin C |
| Warned about purging | ✅ Detailed timeline | ✅ Mentioned | ❌ Not mentioned | ✅ Brief note |
| Introduction schedule | ✅ 4-week plan | ✅ 3-week plan | ❌ "Start slow" | ⚠️ Vague |
| Safety warnings | ✅ SPF mandatory | ✅ SPF + retinoid caution | ⚠️ Generic | ✅ SPF + purging |
What Each AI Recommended
ChatGPT led with adapalene (Differin) as the acne cornerstone — the single most evidence-backed OTC retinoid for acne. Paired it with a gentle hydrating cleanser, azelaic acid for hyperpigmentation, and budget-specific Neutrogena SPF. Included a 4-week introduction schedule so actives don't overwhelm a beginner's skin. Total: $38/month.
Claude took a more conservative approach — started with salicylic acid 2% rather than adapalene, reasoning that a complete beginner should start gentler and escalate if needed. Recommended niacinamide for dual oil control and hyperpigmentation. Notably, Claude was the only AI to mention that hormonal acne at the chin may not respond to topicals alone and suggested discussing with a doctor if no improvement in 8 weeks. Total: $35/month.
Gemini gave the shortest response — a basic cleanser-treat-moisturize-SPF framework without specific product prices or an introduction schedule. Named Neutrogena and CeraVe broadly. Didn't address the hyperpigmentation concern with any targeted ingredient. Useful as a starting point but not actionable without follow-up.
Perplexity provided sourced recommendations with links to articles and product pages. Recommended benzoyl peroxide (with contact therapy method) over retinoids, which is a valid alternative approach. Each recommendation came with a citation. Went slightly over budget at $41/month.
Verdict: Showdown 1
🏆 Winner: ChatGPT — Most complete, most actionable, hit budget target, addressed all concerns with specific products.
🥈 Runner-up: Claude — More cautious but genuinely wise safety callout about hormonal acne requiring professional follow-up.
🥉 Perplexity — Good sourcing, slightly over budget, different but valid approach.
4th: Gemini — Too generic for this specific profile.
Showdown 2: The Rosacea + Anti-Aging Dilemma
Profile
45F, sensitive skin with rosacea type 2 (papulopustular), concerned about first visible wrinkles and loss of firmness. Budget: $120/month. Lives in Scottsdale, AZ (extreme dry heat, intense UV). Currently uses a random "sensitive skin" cleanser and moisturizer, no sunscreen (!!). Tried retinol once — face turned beet red.
Results Summary
| Category | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| Addressed rosacea first | ✅ Priority 1 | ✅ Priority 1 | ⚠️ Equal weight | ✅ Priority 1 |
| Anti-aging approach | Azelaic acid + peptides | Bakuchiol (retinol alt) | ❌ Retinol recommended | Azelaic acid + niacinamide |
| SPF emphasis | ✅ Critical — desert UV | ✅ Detailed — UVB + dry heat | ✅ Mentioned | ✅ Mineral SPF specified |
| Retinol handling | ⚠️ Suggested low-% retry | ✅ Avoided entirely | ❌ Recommended as essential | ✅ Avoided — too risky with rosacea |
| Trigger awareness | ✅ Heat, alcohol, fragrance | ✅ Comprehensive list | ❌ Not addressed | ✅ Sun and heat |
| Dermatologist referral | ✅ For prescription options | ✅ Azelaic acid Rx + laser | ❌ No | ✅ Mentioned |
The Critical Difference
This showdown exposed a safety gap in Gemini. Despite the profile clearly stating rosacea type 2 and a prior retinol disaster, Gemini recommended retinol as "essential for anti-aging" with only a brief "start with low percentage" caveat. For a rosacea patient who already reacted badly, this advice could cause a severe flare.
Claude's approach was the safest and smartest. It recognized the retinol intolerance, pivoted to bakuchiol (a plant-based retinol alternative with emerging evidence and significantly lower irritation risk), and recommended prescription azelaic acid 15% (Finacea) which treats both rosacea AND hyperpigmentation — a two-for-one that addresses both concerns simultaneously.
ChatGPT was good but slightly riskier — it suggested trying retinaldehyde (a gentler retinoid form) at 0.01% concentration, which could work for some rosacea patients but remains a gamble for someone who already reacted.
Perplexity played it safe with a rosacea-first approach and sourced its azelaic acid recommendation to a clinical study.
Verdict: Showdown 2
🏆 Winner: Claude — Superior safety logic, best ingredient pivot (bakuchiol), only AI to recommend specific prescription options.
🥈 Runner-up: ChatGPT — Comprehensive but slightly aggressive with retinoid retry.
🥉 Perplexity — Safe, sourced, but less creative in ingredient alternatives.
4th: Gemini — ⚠️ Potentially harmful recommendation. Retinol for active rosacea is a known contraindication.
Showdown 3: The K-Beauty Enthusiast's Dupe Hunt
Profile
28F, normal-to-dry skin, interested in K-beauty multi-step routine but finding it expensive to import. Top concerns: dullness and early fine lines. Budget: $60/month. Wants equivalent products available from US retailers. Currently using a 10-step imported K-beauty routine costing $140/month.
Results Summary
| Category | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| K-beauty knowledge | ✅ Deep | ✅ Moderate | ⚠️ Surface | ✅ Sourced |
| Dupes identified | 8 product swaps | 5 product swaps | 3 generic swaps | 6 product swaps |
| Savings calculated | ✅ $140→$58/mo | ✅ $140→$62/mo | ❌ No pricing | ✅ $140→$55/mo |
| Maintained routine philosophy | ✅ Kept layering logic | ⚠️ Simplified to 7 steps | ❌ Reduced to 5 steps | ✅ Kept 10-step frame |
| Ingredient match quality | ✅ Active-matched | ✅ Active-matched | ⚠️ Category-matched | ✅ Active-matched |
Best Dupe Finds (Across All AIs)
| Original K-Beauty Product | Price | AI-Found US Dupe | Price | Savings |
|---|---|---|---|---|
| COSRX Advanced Snail Mucin | $25/100ml | CeraVe Moisturizing Cream (different mechanism, similar result) | $16/539g | 74% |
| Sulwhasoo First Care Serum | $90/90ml | Missha Time Revolution Essence (still K-beauty but much cheaper) | $22/150ml | 82% |
| Laneige Water Sleeping Mask | $34/70ml | Neutrogena Hydro Boost Water Gel | $20/50ml | 33% |
| SK-II Essence | $185/160ml | COSRX Galactomyces Essence | $21/100ml | 85% |
| Banila Co Clean It Zero | $22/100ml | Pond's Cold Cream | $8/269g | 72% |
Verdict: Showdown 3
🏆 Winner: ChatGPT — Deepest K-beauty knowledge, most dupe swaps, maintained the 10-step layering philosophy while hitting budget.
🥈 Runner-up: Perplexity — Excellent sourcing with links to where dupes are available.
🥉 Claude — Good quality but simplified the routine more than requested.
4th: Gemini — Didn't engage with K-beauty specifics; gave generic Western alternatives.
Showdown 4: The Ingredient Label Detective
Profile
"I just bought this moisturizer. Here's the ingredient list: Aqua, Glycerin, Cetearyl Alcohol, Caprylic/Capric Triglyceride, Niacinamide, Ceteareth-20, Dimethicone, Tocopheryl Acetate, Panthenol, Allantoin, Phenoxyethanol, Ethylhexylglycerin, Carbomer, Sodium Hydroxide, Parfum. Is it good for my 34-year-old combination skin with large pores?"
Results Summary
| Category | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| Ingredient-by-ingredient analysis | ✅ All 15 | ✅ All 15 | ⚠️ Key 6 only | ✅ All 15 |
| Identified concerns | Parfum (fragrance) | Parfum + Ceteareth-20 | Fragrance noted | Parfum flagged |
| Niacinamide assessment | ✅ "Position 5 = likely 2-5%" | ✅ "Moderate concentration" | ⚠️ "Contains niacinamide" | ✅ "Mid-list = decent %" |
| Suitable for profile? | "Good but not great" | "Adequate, not optimal" | "Yes, suitable" | "Decent but upgradeable" |
| Suggested upgrades | ✅ 3 alternatives | ✅ 3 alternatives | ❌ None | ✅ 2 alternatives |
The Fragrance Debate
All AIs flagged "Parfum" as worth noting, but they disagreed on severity:
- ChatGPT: "Fragrance is the #1 cause of cosmetic contact dermatitis. For combination skin with pore concerns, a fragrance-free version would be better."
- Claude: "Fragrance at the end of the INCI list means low concentration. For non-sensitive skin, this is generally fine. However, if you're using actives (retinol, acids), fragrance adds unnecessary sensitization risk."
- Gemini: "Contains fragrance — some people prefer fragrance-free."
- Perplexity: "Parfum/fragrance is a known irritant for ~8-12% of the population (citing AAD study)."
The nuanced truth: Claude's answer was the most complete. Fragrance at low concentration is fine for most people, but becomes a problem when layered with active ingredients.
Verdict: Showdown 4
🏆 Winner: Claude — Best nuance, most thorough analysis, correctly identified that ingredient position matters for concentration estimation.
🥈 Runner-up: ChatGPT — Complete analysis with actionable alternatives.
🥉 Perplexity — Good sourcing, solid analysis.
4th: Gemini — Incomplete analysis, no alternatives suggested.
Cross-Showdown Patterns
After running these 4 tests (and 6 more not shown here), clear patterns emerged:
When to Use Each AI for Beauty
| If You Need... | Use This AI | Why |
|---|---|---|
| Complete routine with products & prices | ChatGPT | Most specific product knowledge, best at hitting budgets |
| Safety-first approach for sensitive/reactive skin | Claude | Most cautious, best dermatologist-referral logic |
| Quick product price check or availability | Gemini | Real-time shopping integration |
| Sourced recommendations with citations | Perplexity | Links to studies and product pages |
| Ingredient deep dive or INCI analysis | Claude or ChatGPT (tie) | Both excel; Claude is more nuanced, ChatGPT more complete |
| K-beauty or niche brand knowledge | ChatGPT | Broadest training data on niche beauty brands |
| When you have rosacea, eczema, or psoriasis | Claude | Best safety margins for inflammatory conditions |
The Multi-AI Strategy
The smartest approach: ask 2+ AIs the same question, then look for consensus. When ChatGPT, Claude, AND Perplexity all recommend azelaic acid for your concern — that's a strong signal. When they disagree, dig into WHY — the disagreement itself teaches you something about the trade-offs.