Category Archives: Artificial Intelligence

Artificial Intelligence, Bing Image Creator, ChatGPT, Hair Transplant

Fake Hair Transplant Before and After Photos Using AI

18 Comments
Hair Transplant Front Before After AI Photo
Hair transplant in the front (two separate photos). Generated by AI using Bing Image Creator and DALL-E.

Artificial Intelligence (AI) is all the rage in 2023. This is entirely due to the release of the ChatGPT (Chat Generative Pre-trained Transformer) chatbot from OpenAI in November 2022. Most people seem to already be impressed with this technology in its first iteration.

At the time of release, ChatGPT was based on OpenAI’s GPT-3.5 language model and only included information and data through 2021. This was since supplemented with the latest GPT-4 language model that was released on March 14, 2023.

Moreover, after Microsoft Bing added ChatGPT into its new search engine (only available on the Edge browser), the chatbot’s data includes the latest information from 2023. When used inside Bing, the chatbot also cites sources and references that it finds from the Bing search engine.

As the AI chatbot plus search engine battle accelerates, Alphabet (via Google) launched its own new Bard chatbot and Baidu (China) launched Ernie. Google is being more cautious with Bard in comparison to Microsoft with ChatGPT. It would be ironic if the global dangers of AI (and personal data collection) are ultimately in the hands of the Bing, Google and Baidu search engines.

Everything is moving at breakneck speed, although I am not one to trust AI content without detailed verification on my part. On a related note, check out my 2021 post on Alphabet-owned DeepMind’s Alphafold protein database; and my 2022 post on artificial intelligence and drug discovery.

Bing Image Creator Just Launched

I was initially planning to write a post on ChatGPT and how it responds to questions related to hair loss. However, then something even more interesting happened:

On March 21, 2023 Bing launched the groundbreaking Bing Image Creator. It is powered by DALL-E, also from OpenAI.

The old adage is now updated:

“A (fake AI-created) picture is worth a thousand words.”

As Deepfake voice and video artificial intelligence technology continues improving, perhaps someone will create a new adage:

“A (fake AI-created) video is worth a thousand pictures.”

Fake AI Generated Hair Transplant Before and After Photos

Over the years, I have seen numerous examples of hair transplant clinics using before and after patient photos from other clinics as their own. It is an epidemic according to some surgeons. Even when it comes to my own site, I regularly find other spammy sites plagiarizing and scraping my photos, logos, images and content.

But now we have Bing Image Creator that is free to use. No need to plagiarize photos from other people anymore. Moreover, other AI image creation software products (not free) such as Midjourney and Stable Diffusion already became popular in 2022.

The quality of these images is still subpar on some occasions. e.g., the below are two of the four images I got generated for “hair transplant before and after“. It seems like the AI prefers hair transplant results on the rear of the head for some reason! And it often puts the after result first, followed by the before.

Hair Transplant Before After AI Image
Hair transplant before after AI image using Bing Image Creator.

However, I have no doubt whatsoever that it is only a short time before these fake AI generated images become undetectable from the real thing. All of these technologies are constantly learning and self-improving from user feedback.

Moreover, if you vary the query in Bing Image Creator, the photos can become at least somewhat believable. At the very top of this post is one with the query “hair transplant in the front.”

Below are two results for the query “FUE hair transplant.”

FUE Hair Transplant AI Image
FUE hair transplant artificial intelligence Images.

To me, both of the above images seem like they are of real human scalps. Quite amazing if you are someone who is older than 30 years old in my opinion. Google was only founded in 1998, and the internet only started to became popular in the mid-1990s. Smartphones and faster speed internet access are yet more recent.

Inevitably, we will see numerous hair transplant surgeons and hair loss product developers use such “fake” photos. It will become impossible to discern truth from fiction. I would not be surprised if even some clinical trial before and after photos turn out to be doctored ones in the future.

Buyer beware (Caveat Emptor) is truer then ever before.

Artificial Intelligence, Machine Learning

AI and Machine Learning for Hair Loss Drug Discovery

25 Comments

In the past, I briefly discussed the use of artificial intelligence (AI) and machine learning (ML) in drug discovery. Especially when it comes to the potential of rapidly testing new compounds to treat hair loss. Funding for drug discovery startups that use AI is now really taking off.

Update: March 31, 2023 — Speeding up drug discovery with diffusion generative models such as the DiffDock molecular docking model.

Using AI to Predict Hair Loss Compounds

Recently, “Lorence” posted an interesting link to an article that is titled: “Researchers use AI to predict compounds that could neutralize baldness.” The actual study from China is here and it was published on October 20.

In the article, they mention that male pattern hair loss is caused by androgens, inflammation or an overabundance of reactive oxygen species. One potential treatment for the last mentioned is via the creation and utilization of “nanozymes” that mimic the superoxide dismutase (SOD) enzyme. SOD helps fight damaging oxygen free radicals.

The scientists tested machine-learning models with 91 different transition-metal, phosphate and sulfate combinations. A highly efficient manganese thiophosphite (MnPS3) based SOD mimic was discovered using machine learning tools. These ML techniques predicted what cobination would have the most powerful SOD-like ability.

The team subsequently prepared MnPS3 microneedle patches which they used to treat androgenic alopecia-affected mouse models. Microneedling allowed the MnPS3 to penetrate deep layers of the skin (where hair follicle stem cells reside) and remove the excess reactive oxygen species. Within 13 days, the animals regenerated thicker hair strands that more densely covered their previously bald backsides.

Insilico Medicine

Several years ago, there was much hype about a new AI drug discovery company named Insilico Medicine. Among the conditions the company aimed to develop drugs for included hair loss. While they mentioned hair in a number of their past press releases, you no longer see it on their pipeline page.

Neither is it mentioned in a very recent interview with their CEO. The company was covered in the below 2020 video on artificial intelligence for hair loss.

Other Hair Loss Companies using Artificial Intelligence

Also of interest, in late 2019, Iktos (France) and Almirall (Spain) signed an agreement in which Iktos’ AI modelling technology would be used to design novel optimized compounds for Almirall. The latter is a company that is entirely focused on skin and other dermatological conditions. They currently make the world’s only topical finasteride product that has undergone rigorous clinical trials.

Another hair loss company making use of artificial intelligence is South Korea’s Epibiotech (h/t “Ben”). In October 2021, it signed an agreement with CN.AI in order to accelerate the discovery of new hair loss drug candidates.

Opensource Databases and More

Make sure to also read my post on the publicly available DeepMind AlphaFold protein database. Also of interest are open source sites such as the Driskskell Lab’s skin regeneration and wound healing related datasets.

More recently, Dr. Maksim Plikus and his team at UCI developed CellChat, which enables the better understanding of cell-to-cell communication and signaling.

Note that artificial intelligence is also being used for other purposes in the hair loss world. Among these include:

  • Fully automated hair growth detection and measurement systems.
  • New deep learning-based systems used to quantify hair characteristic by scalp area.
  • Tools to help with hair loss self-diagnosis.
  • The ARTAS robotic hair transplant system also uses AI technology when it comes to hair transplants. Only during the graft extraction process for the time being.
  • And as of 2023, the generation of real and fake before and after hair growth photos.