Hey Reader,

There's been a lot of talk about hallucinations this week... and probably not the type you're thinking of. ;)

I'm talking about AI hallucinations. This is when AI makes stuff up.

This happens because generative AI (like ChatGPT and Gemini) is designed to do just that: generate content. It is not designed to research, summarize existing facts, or fact-check. It is designed to create something that matches the prompt and seems plausible.

You can even ask it yourself, and it will be upfront about it.

Why are we talking about AI hallucinations in a health and wellness newsletter?

Well, because there is strong evidence that the recent MAHA report that RFK Jr will be using to suggest policy for the Department of Health was generated at least partially by AI. We know this for a few reasons, including specific tags inserted into references, made-up references to studies that do not exist, and fabricated interpretations of those false studies in the report itself.

Basically, the team cheated and made up studies in this report!

That's dangerous for multiple reasons. If this report were to be used to inform policy, that policy would be based on incorrect information. This could potentially lead to bad policy.

Since this is health policy, the implications could be severe.

Bad health policy can lead to real impacts on people's health and impact their survival rates in the face of multiple health challenges. For example, if, because of this report, the agency recommends against a care protocol, that protocol might no longer be covered by insurance, even if a health provider recommends it. They might also move funding away from research on particular topics towards others based on false pretenses, which could minimize the collective ability to progress with cures.

In other words, lives are on the line. This is why it is so important for these officials to get it right!

What did they do when they got caught?

A student who did this would receive a failing grade and potentially even be kicked out for violating the honor code.

The administration's response was that these were formatting errors, but these are much deeper and more fundamental problems than simple formatting.

A formatting issue is that the citation has the information in the wrong order, is missing punctuation, or has incorrect bolding. Formatting issues are not incorrect or false information. This report is a misrepresentation and falsification of information, not simple formatting errors.

Also, it means those leading the Department of Health are either not understanding the tools they are using, or they are purposely using false information to mislead.

Neither of those is great! We should expect better from the people nominated to lead our health.

Where do we go from here?

I think it starts with demanding better. We need to make it clear that falsifications generated by AI should not be guiding the kind of healthcare we receive in this country. AI is and will be a part of the landscape, but it requires due diligence and verification. It is a tool to be used by the professionals, not to replace them. We should ask for health recommendations and policy decisions to be guided by professionals who evaluate the data, perform inquiries, and complete the due diligence of research to make informed recommendations to improve our healthcare.

For me, it calls into question the entire report. How are we to reliably trust the content of the report knowing that these major issues with the integrity of the content exist?

It also means working together! I'm so happy you're here to have these conversations with me and work together to build a better system and better health for ourselves and each other. When things get challenging, I like to say we are living in interesting times. This moment is definitely interesting in terms of the future of health policy and programs in this country.

Is this report and the hallucinations something you want to hear more about? Just reply if that's the case, and I can make a longer post.

I also think you're going to love this week's recipe! It's super flexible, meaning it can work with lots of different fresh produce for different takes.

Non-existent studies cited in the MAHA report The original report has since been removed, but the original report cited false studies that the authors confirmed were never written. Read article

One researcher's attempt with AI reveals multiple falsifications These hallucinations are well known in the scientific community, and this article shares one particular venture as well as references some stats on the frequency. (Note: I generated this image with AI) Read article

Now, for the tasty treats

Recipe Spotlight

PS. When you are ready to take your journey a little further, here are 2 ways I can help:

1. Start journaling your own health goals and needs to better understand the language of your body and how to best support your bio-individual health needs with the Wellness Journal.

2. Personalized coaching allows us to work one-on-one to evaluate your particular health struggles and goals. We'll work to understand your unique needs and put together a plan that works to create a life full of vitality and joy. Schedule your FREE intro call!

​

You are receiving this email because you signed up for emails with Eat Your Way Clean! This newsletter is monthly. If you loved this information, spread the word to your friends who can use this link to sign up!​

113 Cherry St #92768, Seattle, WA 98104
​Unsubscribe · Preferences​