OK ChatGPT, suggest some tweaks for me!

Have you thought of asking your favourite artificial-intelligence chatbot to critique your face and suggest appropriate tweakments? No, neither had I, until practitioners began telling me that patients are coming in with a laundry-list of procedures they think they need because… ChatGPT et al said so, and they don’t like being told they’re asking for the impossible, or for treatments that are inappropriate or over-the-top. It’s becoming a serious issue for responsible clinics. Yet, naturally, once I heard this was happening, I felt compelled to load up some photos and see what AI would suggest for my own face…

What are people asking AI for?

“Anything and everything,” says Rosey Drewitt-Staples, a leading nurse-practitioner and aesthetics trainer based in Nottingham, and they’re creating a fair old headache for clinics. When she gently explained to one young patient that the procedures he was requesting – straight from an AI generated treatment plan – weren’t appropriate and wouldn’t create the look he wanted, he became very cross, wouldn’t accept that his expectations were unrealistic and left the clinic. Oh dear.

“I’m certainly seeing an increase in patients bringing lists from AI or using AI-edited images,” says Dr Ed Robinson. “The vast majority of them have been ridiculous or so non-specific that they are meaningless. One or two have made sensible suggestions but the images generated have never been realistic. Under-eye filler is commonly suggested  – though in reality, in my experience, it is very rarely the answer. I had a patient a couple of days ago who had been told they needed every treatment I have in my portfolio, which didn’t do wonders for their self-esteem.”

“Yes, lots of people are doing this,” agrees Alexandra Mills, an elite nurse practitioner and vice-chair of the British Association of Medical Aesthetic Nurses (BAMAN). Is it a bit like taking pictures of celeb haircuts to the hairdresser and wanting to look the same? “Yes indeed, but worse!” she replies.

What I asked ChatGPT

Fully expecting nonsense, I uploaded my photos – taken bare-faced, in strong daylight. “I want you to act as an elite, experienced and direct-talking aesthetic practitioner,” I typed, “and after studying these photos, suggest what tweakments would be helpful to enhance my face.” I took a deep breath and braced myself for the result. A few seconds later I had my plan. Here, it announced, was a professional assessment based only on what it could see in the photos, “but written with the same directness and nuance you would get from a highly experienced practitioner who understands your face, your history, and your aesthetic preferences. Importantly, this is not prescriptive and is purely informational.” Wow. So far so thoughtful.

After a few flattering comments (LLMs are good at those, unless you tell them not to), the main message was that what my face needed was fine tuning, refinement and support – rather than looking to change anything.

The treatments AI suggested for me

Far from being way off-beam, the suggestions struck me as very sensible. In my mid-face, it spotted the flattening that happens in the lateral cheeks with age and suggested small boluses of HA filler “not to inflate, just to maintain projection”, and skinboosters or polynucleotides for skin quality and elasticity. For my lower face, it picked up the “age-typical mild soft-tissue descent towards the marionette zone” – that sag down towards the chin that we all hate – and suggested tiny bits of filler “buttresses” in the pre-jowl sulcus (the dent in the jawline where it meets the chin) if I felt the area needed firming. (Yes, I do!), and toxin to reduce the DAO (the depressor anguli oris, aka the grumpy-face muscle that pulls down the corners of the mouth, which is something I have done regularly.)

For my neck, generative treatments like RF microneedling for collagen and skin-thickness, hydrating skinboosters and polynucleotides were the suggested options, and for the slight hollowing around my eyes, more skinboosters or polynucleotides or tear trough filler only if there was a true volume deficit (there isn’t, so no thanks).

To keep up my skin quality, the suggestions were light fractional laser once or twice a year, maybe a gentle peel or two, more RF microneedling and regular doses of polynucleotides.

Are these AI tweaks right for me?

To me this seemed pretty good – albeit quite a lot of treatment, and quite expensive…

“This doesn’t sound unreasonable,” says Dr Robinson. “The focus is on regeneration and very subtle volumising in anatomically sensible places. The plan often says, “if we’re seeing x, then a certain treatment is suitable. This is where an assessment from an aesthetic expert, face-to-face, in 3D and with movement, from all angles, comes into play. And where the plan suggests a variety of options, to stimulate collagen, or maintain skin quality, this is where you need real-life expertise to look at your medical history, your lifestyle, your budget, and whether you’re okay with downtime, and tailor a treatment plan to suit you. The trouble with this sort of AI-generated programme is that you could always take that plan to an inexperienced practitioner who might try to replicate it to the letter and that could cause potential harm.”

Alexandra Mills echoes this. “You need to see the patient to assess their movement-symmetry, and touch their skin and feel it’s quality, see how they smile… the list goes on.”

So, what would I look like?

It had been going so well. Then I asked the chatbot to show me how I might look after all these treatments. The best I can say is: wildly unrealistic. “There’s a miscommunication here,” says Drewitt-Staples. “They’ve changed you into something attractive but unrecognisable. Any experienced practitioner knows that that image, with that robotic, hologramic look, can’t be achieved with the suggested treatments. Also, it has given you cheek projection and lip enhancement that isn’t mentioned in the plan! Unfortunately, younger patients might not realise that AI images are giving a false impression of what’s possible, and it’s feeding unrealistic expectations.

Dr Robinson echoes this: “For the most part, the treatment plan is sensible but it’s not going to make you look like these, filtered, flawless images. It talks about ‘subtle results’ but it has also changed your lips and eyebrows, so it doesn’t really look like you at all.”

The “afters” images have also made my face much more symmetrical. This sounds like a good idea, until you take a look at AI-optimised images of, say, the Mona Lisa or Timothée Chalamet. Humans are not summetrical, that’s part of what makes us look real.

“There is no shortcut to great results,” advises Drewitt-Staples. “It still comes down to the human eye which appreciates the beautiful nuances of asymmetry – and which we’re not getting here. A face that’s too symmetrical looks unnervingly unreal to other people, and creates distrust. We are entering a challenging arena in aesthetics where we have to factor these kind of images into the way we discuss treatments with patients.”

The verdict

As a rough, high-level analysis, my AI tweakment plan was surprisingly sensible. But the projected “after” images were pure fantasy and therein lies the danger.

While AI can outline possibilities based on what it can see, only a trained practitioner can assess all the contributing real-life factors of bone structure, facial movement, skin quality, personality, lifestyle, medical history, and pair this up with where tweakments can realistically help.

What to find out more about the best tweakments and skincare for you? Head to The Guide for more advice and articles.