As someone whose life’s work was stolen to create ChatGPT and who is very against the negative effects AI has on health and humanity, I’m so disappointed to hear this. Maybe you can allay my fears…how will you be building this, what will it be trained on, will you respect copyright, and how do you justify the environmental harm, and thus the harm to humanity and all living beings, done by these tools? How can we be healthy on a planet that is boiling away all its water on chatbots?
Hi Molly- Thanks for your comment. I'm so sorry your work has been stolen. I have been moving very slowly and cautiously on this project, because we share a lot of your concerns. I decided to work with Dewey, specifically, because it's a closed model that is trained on YLE content only. If "Ask YLE" doesn't find the answer in our archives, the chatbot won't answer it. (You can think of it is as a librarian or smarter search tool) The "Ask YLE" tool will also help me keep track of questions that we don't have an answer to yet, so that the team and I can try to answer them in the future. I am looking for help from the YLE community to test this tool, because we want to make sure that it actually helps you find information from our archive.
Maybe the Dewey team can give you estimates to share about how big the training data set is (how many bytes) and how much power will be consumed training the AI. I'm guessing the numbers will be much smaller than for a general-purpose AI like ChatGPT or Gemini. (And I assume the training data will not include the comments section, correct?)
I understand and thank you for the reply. I know accurate scientific communication is a passion of yours which is why I’m scratching my head over this. We all know how confidently AIs reply to questions, with low levels of being actually correct, as well as their people-pleasing bias that leads to other problems. This seems against your mission, your goals, and your ethics. (And the impact on human health and wellness as the planet warms did not figure into your reply.) I have been a paid member for about a year and I should tell you this has me reconsidering my membership. I love supporting you and your work and I think you are amazing but I can’t in good conscience fund more earth-warming lie machines. Just my .02 (maybe a little more than two cents but this is a huge concern of mine, and I wish it was a huge concern of everyone else’s).
I have two concerns. First, my understanding of the AI's job description here is to find what YLE and guests have written about a particular subject. Material in the comments section sometimes reflects views that YLE disagrees with. Second, when I read the Terms of Use for Substack, it's not clear to me that YLE would have a legal right to use comments for AI training.
That's correct!! Nothing in the comments sections. Just previous posts, PDF's I've written, the podcast content that I was on, and presentations I make. That's it! I can definitely ask about the bytes and how much power will be consumed. I don't know that answer but can try and find it.
If helpful for the power question, I talked to a friend who has worked in programming LLMs, and my understanding from talking to her is that power usage is sort of three questions:
1. How much power was used to generate the broader language model that your data is being integrated into.
2. How much power is used to add your data to the model.
3. How much power is used to generate the individual responses.
My sense is that 2 and 3 *might* be small, but 1 is likely to be quite large.
If we are to follow the vaccination programs of countries like Denmark, then we should have the universal healthcare from birth to grave as well as a broad primary care system that provides access to everyone that citizens in countries like Denmark enjoy.
Please reconsider the chatbot idea. What problem are you actually trying to solve with a chatbot? I really believe the many negatives outweigh the positives.
Hi Wayne. Thanks for your question. We get a LOT of emails and comments from people asking us questions, and we're a small team. We think a tool like this will help people find answers we've provided in the past, and help us discover important questions that we can tackle in the future. This isn't a chatbot that will provide extensive, personalized advice; it's more of a librarian or search tool.
I will be the first to admit I am no expert. (My wife will be the second, or maybe the first, depends on the day.) What about a roll your own search engine? Just searches your data. Or, maybe a self-hosted AI trained on your data only. This might sidestep the water/electrical/copyright issue a bit as you skip the data center and use only your data. What if all your archived data was located somewhere on the web and could be searched using a link to a directed search (like a bang search maybe). Might require training us users a bit.
I have to say that you and your team are doing a hell of a job. These ideas may require skills and or time your team does not currently have and I am loathe to suggest anything that winds up slowing you down.
Your suggestions don't slow us down! In fact, they help an enormous amount, so thank you. I've found the Substack archive search engine TERRIBLE. I've been so frusterated with it, i can't even find things I know I've written about before. We will look into the bytes and energy, that's an important point.
Given medical research’s often blind eye to women and the recent Stanford study noting AI bias against women https://news.stanford.edu/stories/2025/10/ai-llms-age-bias-older-working-women-research how is your chatbot going to protect women in search of answers? I know there are pitfalls with Dr. Google, but unlike others who dismissed my 95-year-old mother’s hallucinations as dementia, her PCP credited me for the quality of my internet research determining the problem to be Charles Bonnet Syndrome suffered by a third of those with dry-eye macular degeneration and for finding a British study on the aged showing a positive impact of gabapentin on aged sufferers—so prescribed her the low-dose gabapentin that relieved her hallucinations. In short, I would rather frame my own questions and evaluate the reliability of sources than trust a chatbot.
As an 80-year old asthmatic, I regularly get all my vaccinations, especially COVID-19 and flu (yes I got my RSV a couple of years ago). The thing I fear most is being unable to get COVID vaccines and ending up in the hospital with even more difficulty breathing. I wear a mask everywhere I go most of the year. I take a break when grocery shopping in the summer, but that's it. The weird thing to me is seeing how few people wear masks.
"The weird thing to me is seeing how few people wear masks."
Same! I haven't had a cold, covid, flu, or anything else in five years, and I used to get them all the time. Even if we eliminated covid, I'd still be masking, both for me and for others.
tbh, these days I think of not masking in places like public transit as a lot like parking in a disabled parking spot. If my unmasked presence somewhere means an immunocompromised person can't safely be there, then I'm participating in forcing them out of that space, plain and simple, just as if I were parking in a disabled spot.
I hadn't thought about immunocompromised people, so good on you for reminding me of that. Hopefully anyone who is immunocompromised is smart enough to wear a mask always. I find masks hard to breathe in, which is why a break when I shop for groceries in the summer is helpful. The other thing about wearing a mask is that I have coughing asthma, so my mouth and nose are automatically covered if I have an attack.
I was interested in the list of things educators and clinicians can do (I'm a retired educator) but I live in Denver - I don't want to pay for a subscription to the NY newsletter to see it. Is this information available anywhere else?
As to the question about chatbots - as an educated person who is not a medical professional, I MAY (not will, just may) use such a tool to generate questions I want to research further, but as a cancer survivor who participates in several support groups, I have seen an increase in people using such tools and reporting the results as fact. I find this incredibly concerning.
Thanks! I went straight to it from the link in the article... it's still asking for a subscription to the NY newsletter... Can you put a direct link here, maybe?
My son took his recently-turned 5 year old to an annual pediatric wellness visit at Kaiser in Southern California. My son requested a Covid-19 vaccination for my grandson, and my son also expected he himself could be directed to a clinic there at Kaiser where he could receive the Covid-19 vaccine, as he has in every prior year since Covid-19 immunizations became available. Instead, the Kaiser
physician sighed as his chin dropped to his chest, and he made comments that led my son to believe he would have to forcefully push back against that provider in order to receive the vaccinations. My son and grandson left the appointment without the desired immunizations. I am confused. I thought physicians in California would follow guidance from our new west coast states health collective.
When was that, exactly? I had to delay my COVID vaccine from SoCal Kaiser until early October, due to ACIP delayed approval. I had no difficulties in getting an appointment, but I do have proven risk factors for more severe infection. (Alas, I did have to delay my vaccine as I got COVID for the first time a week before my appointment. Thankfully, no complications.) Perhaps KP was waiting for clarification about the Western Alliance, or the physician wasn't updated on this information. I hope your son will try again asap.
Ohhhh I wonder if this is contributing to the rates. The big ACIP delay and the implications from that. I hope we catch up from that. I will be watching the data closely.
Cherie, thank you. My son and grandson were at SoCal Kaiser right before Thanksgiving. I think perhaps you are right that the particular physician was not updated on the Western Alliance recommendations. Also, neither of them has underlying conditions. However, my daughter-in-law was not given the vaccine either. Their older son did receive it, because he has asthma.
Cherie, that is a great idea. He asked me not to pester him with more information about this year’s Covid-19 shot, but I think I may risk being a pest and pass along your suggestion.
Regarding, flu like illnesses; In the Sacramento area, we may be witnessing an outbreak of Mycoplasma pneumoniae infections with some resulting in "walking peumonia". An outbeak has occurred in a local school where several children have been diagnosed with pneumonia. The school officials have sent out a notice/request alerting parents of what has happened while asking parents to not send their children to school with a cough.
I'm adding a bit for non-clinicians. Mycoplasma pneumonia is a bacterial infection of the lungs that is transmitted person to person through air droplets (coughing, sneezing, breathing) or by surface contact, so handwashing, well-fitting masks, avoiding crowds, etc. are basic and effective prevention. It's usually relatively mild (hence, the tag "walking pneumonia"), but it can be severe and cause complications outside the lungs. Young children, older adults, and others with weaker immune systems are especially vulnerable. Because Mycoplasma pneumoniae bacteria paralyze the cilia that sweep mucous out of the airway, it frequently causes a nagging cough. Most cases can be successfully treated with antibiotics such as azithromycin (Zithromax).
If you do make a chatbot (and I hope you will not for both ethical and practical reasons), please place enormous, obvious, repeated warnings on it that nothing it says should be taken at face value without double-checking against legitimate sources. Factual errors are mathematically inevitable with all language learning models (link: https://www.computerworld.com/article/4059383/openai-admits-ai-hallucinations-are-mathematically-inevitable-not-just-engineering-flaws.html ). For a trivial practical example, see the discourse on asking ChatGPT how many times the letter B appears in the word "blueberry" (link: https://www.distractify.com/p/chatgpt5-blueberry-results ). The bot will "lie", frequently and confidently, and people will need to be reminded of that every time they use it.
All three of us got Covid three weeks after getting the vaccine. Granted, it seemed like a "light case". More like a cold than full blown Covid but it was so disappointing to see that positive line show up on the tests we all took. Was it just not enough time for our systems to be "loaded"?
The SARS-CoV-2 vaccine (and all vaccines for that matter) does not have a reasonable expectation to prevent “infection”. That is a superpower that current vaccines do not have. Vaccines are effective because they lower the risk of complications like hospitalization and death from the “disease” caused by the infection. You had mild COVID-19 disease caused by the SARS-CoV-2 infection. The vaccine performed remarkably well!
I do wonder if part of the low Covid-vaccine takeup is because those who care enough to take it each season are watching Covid rates, seeing that they are extremely low, and holding off until they show signs of stirring. That's certainly the case in our family.
Question about COVID vaccine: I can’t seem to find a comparable goal for coverage like exists for flu (HP2030). What would you recommend for a target for the general population?
There isn't a target and I wish tehre was too. These days I view the covid vaccine as the flu vaccine, so would want the same coverage, especially for older adults.
The H3N2 strain in the vaccine (which apparently is not very protective against the circulating strain) is called District of Columbia, which seems apt. Who picks the names?
The vaccine adoption numbers this year are bleak. I’m not as concerned about the lack of SARS-CoV-2 vaccine uptake as I am the seasonal influenza vaccine rates. Anyone who has had three or more mRNA SARS-CoV-2 vaccines has more durable protection against hospitalization and death from COVID-19 disease than less robustly vaccinated people with similar age and health conditions.
I’m concerned about your use of the word “waning” in the context of durable immunity. We can measure antibody levels as a marker of humoral immunity. Those levels decrease with time but it’s inappropriate to use the word waning to describe that decrease. Instead, it’s a phenomenon of “contraction” of measurable markers of humoral immunity. The T-cell mediated cellular immunity is preserved to a greater degree than the antibody level would suggest. Infection with SARS-CoV-2 triggers activation of both humoral and cellular immunity to address the infection and reduce the risk of hospitalization and death from COVID-19 disease.
It’s difficult to measure cellular immunity but it’s certain that cellular immunity becomes less robust with time since the most recent vaccine. For that reason I still encourage everyone to get yearly SARS-CoV-2 booster vaccines. And for Pete’s sake, if you haven’t been vaccinated to protect against COVID-19 disease complications, GET VACCINATED. However the consequences of foregoing vaccine boosters are much less dire than failure to get the seasonal influenza vaccine this year.
As someone whose life’s work was stolen to create ChatGPT and who is very against the negative effects AI has on health and humanity, I’m so disappointed to hear this. Maybe you can allay my fears…how will you be building this, what will it be trained on, will you respect copyright, and how do you justify the environmental harm, and thus the harm to humanity and all living beings, done by these tools? How can we be healthy on a planet that is boiling away all its water on chatbots?
Hi Molly- Thanks for your comment. I'm so sorry your work has been stolen. I have been moving very slowly and cautiously on this project, because we share a lot of your concerns. I decided to work with Dewey, specifically, because it's a closed model that is trained on YLE content only. If "Ask YLE" doesn't find the answer in our archives, the chatbot won't answer it. (You can think of it is as a librarian or smarter search tool) The "Ask YLE" tool will also help me keep track of questions that we don't have an answer to yet, so that the team and I can try to answer them in the future. I am looking for help from the YLE community to test this tool, because we want to make sure that it actually helps you find information from our archive.
Maybe the Dewey team can give you estimates to share about how big the training data set is (how many bytes) and how much power will be consumed training the AI. I'm guessing the numbers will be much smaller than for a general-purpose AI like ChatGPT or Gemini. (And I assume the training data will not include the comments section, correct?)
I understand and thank you for the reply. I know accurate scientific communication is a passion of yours which is why I’m scratching my head over this. We all know how confidently AIs reply to questions, with low levels of being actually correct, as well as their people-pleasing bias that leads to other problems. This seems against your mission, your goals, and your ethics. (And the impact on human health and wellness as the planet warms did not figure into your reply.) I have been a paid member for about a year and I should tell you this has me reconsidering my membership. I love supporting you and your work and I think you are amazing but I can’t in good conscience fund more earth-warming lie machines. Just my .02 (maybe a little more than two cents but this is a huge concern of mine, and I wish it was a huge concern of everyone else’s).
Mike, may I ask what concerns you have about the comments section?
I have two concerns. First, my understanding of the AI's job description here is to find what YLE and guests have written about a particular subject. Material in the comments section sometimes reflects views that YLE disagrees with. Second, when I read the Terms of Use for Substack, it's not clear to me that YLE would have a legal right to use comments for AI training.
That's correct!! Nothing in the comments sections. Just previous posts, PDF's I've written, the podcast content that I was on, and presentations I make. That's it! I can definitely ask about the bytes and how much power will be consumed. I don't know that answer but can try and find it.
If helpful for the power question, I talked to a friend who has worked in programming LLMs, and my understanding from talking to her is that power usage is sort of three questions:
1. How much power was used to generate the broader language model that your data is being integrated into.
2. How much power is used to add your data to the model.
3. How much power is used to generate the individual responses.
My sense is that 2 and 3 *might* be small, but 1 is likely to be quite large.
If we are to follow the vaccination programs of countries like Denmark, then we should have the universal healthcare from birth to grave as well as a broad primary care system that provides access to everyone that citizens in countries like Denmark enjoy.
Hear hear!
(Air horn sound effects!)
Please reconsider the chatbot idea. What problem are you actually trying to solve with a chatbot? I really believe the many negatives outweigh the positives.
Hi Wayne. Thanks for your question. We get a LOT of emails and comments from people asking us questions, and we're a small team. We think a tool like this will help people find answers we've provided in the past, and help us discover important questions that we can tackle in the future. This isn't a chatbot that will provide extensive, personalized advice; it's more of a librarian or search tool.
I will be the first to admit I am no expert. (My wife will be the second, or maybe the first, depends on the day.) What about a roll your own search engine? Just searches your data. Or, maybe a self-hosted AI trained on your data only. This might sidestep the water/electrical/copyright issue a bit as you skip the data center and use only your data. What if all your archived data was located somewhere on the web and could be searched using a link to a directed search (like a bang search maybe). Might require training us users a bit.
I have to say that you and your team are doing a hell of a job. These ideas may require skills and or time your team does not currently have and I am loathe to suggest anything that winds up slowing you down.
Thank you all for all you do.
Your suggestions don't slow us down! In fact, they help an enormous amount, so thank you. I've found the Substack archive search engine TERRIBLE. I've been so frusterated with it, i can't even find things I know I've written about before. We will look into the bytes and energy, that's an important point.
Given medical research’s often blind eye to women and the recent Stanford study noting AI bias against women https://news.stanford.edu/stories/2025/10/ai-llms-age-bias-older-working-women-research how is your chatbot going to protect women in search of answers? I know there are pitfalls with Dr. Google, but unlike others who dismissed my 95-year-old mother’s hallucinations as dementia, her PCP credited me for the quality of my internet research determining the problem to be Charles Bonnet Syndrome suffered by a third of those with dry-eye macular degeneration and for finding a British study on the aged showing a positive impact of gabapentin on aged sufferers—so prescribed her the low-dose gabapentin that relieved her hallucinations. In short, I would rather frame my own questions and evaluate the reliability of sources than trust a chatbot.
As an 80-year old asthmatic, I regularly get all my vaccinations, especially COVID-19 and flu (yes I got my RSV a couple of years ago). The thing I fear most is being unable to get COVID vaccines and ending up in the hospital with even more difficulty breathing. I wear a mask everywhere I go most of the year. I take a break when grocery shopping in the summer, but that's it. The weird thing to me is seeing how few people wear masks.
"The weird thing to me is seeing how few people wear masks."
Same! I haven't had a cold, covid, flu, or anything else in five years, and I used to get them all the time. Even if we eliminated covid, I'd still be masking, both for me and for others.
tbh, these days I think of not masking in places like public transit as a lot like parking in a disabled parking spot. If my unmasked presence somewhere means an immunocompromised person can't safely be there, then I'm participating in forcing them out of that space, plain and simple, just as if I were parking in a disabled spot.
I hadn't thought about immunocompromised people, so good on you for reminding me of that. Hopefully anyone who is immunocompromised is smart enough to wear a mask always. I find masks hard to breathe in, which is why a break when I shop for groceries in the summer is helpful. The other thing about wearing a mask is that I have coughing asthma, so my mouth and nose are automatically covered if I have an attack.
I was interested in the list of things educators and clinicians can do (I'm a retired educator) but I live in Denver - I don't want to pay for a subscription to the NY newsletter to see it. Is this information available anywhere else?
As to the question about chatbots - as an educated person who is not a medical professional, I MAY (not will, just may) use such a tool to generate questions I want to research further, but as a cancer survivor who participates in several support groups, I have seen an increase in people using such tools and reporting the results as fact. I find this incredibly concerning.
Hi Karen! I just removed the paywall from that YLE NY post!
Thanks! I went straight to it from the link in the article... it's still asking for a subscription to the NY newsletter... Can you put a direct link here, maybe?
Oh gosh, I totally did that wrong yesterday. It's updated now. So sorry about that!
Thanks so much - it's working now!
Is the flu vaccine protective against the strain that mention keeping your eyes on?
It will be helpful and still important to get, but it will not work as well as we want it to.
My son took his recently-turned 5 year old to an annual pediatric wellness visit at Kaiser in Southern California. My son requested a Covid-19 vaccination for my grandson, and my son also expected he himself could be directed to a clinic there at Kaiser where he could receive the Covid-19 vaccine, as he has in every prior year since Covid-19 immunizations became available. Instead, the Kaiser
physician sighed as his chin dropped to his chest, and he made comments that led my son to believe he would have to forcefully push back against that provider in order to receive the vaccinations. My son and grandson left the appointment without the desired immunizations. I am confused. I thought physicians in California would follow guidance from our new west coast states health collective.
When was that, exactly? I had to delay my COVID vaccine from SoCal Kaiser until early October, due to ACIP delayed approval. I had no difficulties in getting an appointment, but I do have proven risk factors for more severe infection. (Alas, I did have to delay my vaccine as I got COVID for the first time a week before my appointment. Thankfully, no complications.) Perhaps KP was waiting for clarification about the Western Alliance, or the physician wasn't updated on this information. I hope your son will try again asap.
Ohhhh I wonder if this is contributing to the rates. The big ACIP delay and the implications from that. I hope we catch up from that. I will be watching the data closely.
Cherie, thank you. My son and grandson were at SoCal Kaiser right before Thanksgiving. I think perhaps you are right that the particular physician was not updated on the Western Alliance recommendations. Also, neither of them has underlying conditions. However, my daughter-in-law was not given the vaccine either. Their older son did receive it, because he has asthma.
Maybe he could try making a COVID vaccine appointment online, through the KP app.
Cherie, that is a great idea. He asked me not to pester him with more information about this year’s Covid-19 shot, but I think I may risk being a pest and pass along your suggestion.
Regarding, flu like illnesses; In the Sacramento area, we may be witnessing an outbreak of Mycoplasma pneumoniae infections with some resulting in "walking peumonia". An outbeak has occurred in a local school where several children have been diagnosed with pneumonia. The school officials have sent out a notice/request alerting parents of what has happened while asking parents to not send their children to school with a cough.
I'm adding a bit for non-clinicians. Mycoplasma pneumonia is a bacterial infection of the lungs that is transmitted person to person through air droplets (coughing, sneezing, breathing) or by surface contact, so handwashing, well-fitting masks, avoiding crowds, etc. are basic and effective prevention. It's usually relatively mild (hence, the tag "walking pneumonia"), but it can be severe and cause complications outside the lungs. Young children, older adults, and others with weaker immune systems are especially vulnerable. Because Mycoplasma pneumoniae bacteria paralyze the cilia that sweep mucous out of the airway, it frequently causes a nagging cough. Most cases can be successfully treated with antibiotics such as azithromycin (Zithromax).
Thank you Dr. Mandell for providing excellent and appropriate additional information.
Dr. Apgar, thank you for this! I shared with our CA YLE, Matt.
If you do make a chatbot (and I hope you will not for both ethical and practical reasons), please place enormous, obvious, repeated warnings on it that nothing it says should be taken at face value without double-checking against legitimate sources. Factual errors are mathematically inevitable with all language learning models (link: https://www.computerworld.com/article/4059383/openai-admits-ai-hallucinations-are-mathematically-inevitable-not-just-engineering-flaws.html ). For a trivial practical example, see the discourse on asking ChatGPT how many times the letter B appears in the word "blueberry" (link: https://www.distractify.com/p/chatgpt5-blueberry-results ). The bot will "lie", frequently and confidently, and people will need to be reminded of that every time they use it.
See the AI data center poisoning of South Memphis' air for one of the many, many ethical concerns (link: https://www.politico.com/news/2025/05/06/elon-musk-xai-memphis-gas-turbines-air-pollution-permits-00317582 ).
All three of us got Covid three weeks after getting the vaccine. Granted, it seemed like a "light case". More like a cold than full blown Covid but it was so disappointing to see that positive line show up on the tests we all took. Was it just not enough time for our systems to be "loaded"?
The SARS-CoV-2 vaccine (and all vaccines for that matter) does not have a reasonable expectation to prevent “infection”. That is a superpower that current vaccines do not have. Vaccines are effective because they lower the risk of complications like hospitalization and death from the “disease” caused by the infection. You had mild COVID-19 disease caused by the SARS-CoV-2 infection. The vaccine performed remarkably well!
Thank you for your explanation. I figured as much, but you explained it a bit better than the nurse at my clinic.
I do wonder if part of the low Covid-vaccine takeup is because those who care enough to take it each season are watching Covid rates, seeing that they are extremely low, and holding off until they show signs of stirring. That's certainly the case in our family.
Yes, this was a thought I had too! We usually see a bump in vaccinations when the waves starts. Although that would have happened last year too.
Perhaps last year it didn't because of the fairly large summer wave, which we didn't really have this year?
Question about COVID vaccine: I can’t seem to find a comparable goal for coverage like exists for flu (HP2030). What would you recommend for a target for the general population?
There isn't a target and I wish tehre was too. These days I view the covid vaccine as the flu vaccine, so would want the same coverage, especially for older adults.
The H3N2 strain in the vaccine (which apparently is not very protective against the circulating strain) is called District of Columbia, which seems apt. Who picks the names?
(I'm referring to the weekly report for Maine, where I live: https://www.maine.gov/dhhs/mecdc/sites/maine.gov.dhhs.mecdc/files/MaineILIweek2548.pdf)
The vaccine adoption numbers this year are bleak. I’m not as concerned about the lack of SARS-CoV-2 vaccine uptake as I am the seasonal influenza vaccine rates. Anyone who has had three or more mRNA SARS-CoV-2 vaccines has more durable protection against hospitalization and death from COVID-19 disease than less robustly vaccinated people with similar age and health conditions.
I’m concerned about your use of the word “waning” in the context of durable immunity. We can measure antibody levels as a marker of humoral immunity. Those levels decrease with time but it’s inappropriate to use the word waning to describe that decrease. Instead, it’s a phenomenon of “contraction” of measurable markers of humoral immunity. The T-cell mediated cellular immunity is preserved to a greater degree than the antibody level would suggest. Infection with SARS-CoV-2 triggers activation of both humoral and cellular immunity to address the infection and reduce the risk of hospitalization and death from COVID-19 disease.
It’s difficult to measure cellular immunity but it’s certain that cellular immunity becomes less robust with time since the most recent vaccine. For that reason I still encourage everyone to get yearly SARS-CoV-2 booster vaccines. And for Pete’s sake, if you haven’t been vaccinated to protect against COVID-19 disease complications, GET VACCINATED. However the consequences of foregoing vaccine boosters are much less dire than failure to get the seasonal influenza vaccine this year.
In addition, the measles epidemic is growing across the U.S. --
https://gizmodo.com/u-s-measles-outbreak-nears-grim-milestone-as-hundreds-quarantine-in-south-carolina-2000697900