• We have updated Tinnitus Talk.

    If you come across any issues, please use our contact form to get in touch.

Frequency Therapeutics — Hearing Loss Regeneration

Executing mathematical equations and comparing biochemical reactions within the human body are entirely unequal.

Anybody that thinks AI will be solving complex medical research within the next <100 years is living in an episode of Star Trek.
Equations or chemical reactions is irrelevant to AI, all chemical reactions are essentially mathematical equations. Pretty much everything about chemistry is math, it's all an energy and mass balance equation. Conservation of mass tells us the number of atomic species on the left side of a reaction must equally what's on the right, it is literally a mathematical mass balance. Once we feed an artificial model a large dataset of known reactions and train the model accordingly, AI will be much better at predicting what's on the product side than humans. I personally see one of AI's biggest advantages is being able to develop correlations between large sets of research data.

We live in a data driven world, we just need help to better analyze the data we have.
 
It already is since a while back.

Artificial Intelligence: How is It Changing Medical Sciences and Its Future?

And there's tons more examples. Just use Google.
Not a single example in there is comparable to what people are hoping for here.

Those are input/output models, i.e. sort skin cancer images into negative and positives based on a database of each and allow the model to make a choice. If you think that's comparable to replacing researchers trying to figure out the brain, we have wildly different ideas of what AI represents.

I mean, one of the examples was directing someone to the right physician based on questions. FFS, any customer service live chat does that.
Again, nothing in there that is true AI in the sense of what people are hoping here.

AlphaFold is probably the most impressive but it's still nowhere near what people in this thread are hoping for.

You can 'just use Google' all you want, but there still zero tangible evidence REAL AI will be making discoveries for conditions like tinnitus in the next 100 years. All of the examples given were understood by the researchers and an intelligent model was used to aid processing that already know algorithm. Which is used by my lab all the time, and I'd hardly say it even scratches the surface of being AI.

There is no AI that will be able to solve conditions like tinnitus. Sorry, I just don't buy it. Think this is the point where hope becomes desperation. AI is in its infancy, and whilst it may be possible in the future, it may also be possible for us to go to Mars in the future, just not right now.
 
The link you provided said:
ChatGPT's mathematical abilities are significantly below those of an average mathematics graduate student. Our results show that ChatGPT often understands the question but fails to provide correct solutions. Hence, if your goal is to use it to pass a university exam, you would be better off copying from your average peer!
Still nowhere what people in here are hoping for.
 
Still much more than "executing mathematical equations", whatever that means.
And completely irrelevant to tinnitus, or any clinical research. People are hyped about AI because ChatGPT can write a poem or MidJourney can generate images. Extremely mindblowing stuff, especially MidJourney, but be realistic. Of course all new technology develops and grows, but it's decades off assisting in clinical research in any real capacity. I've never said AI will never be at the level of "computer, solve this issue" - but probably not in our lifetime.
Equations or chemical reactions is irrelevant to AI, all chemical reactions are essentially mathematical equations. Pretty much everything about chemistry is math, it's all an energy and mass balance equation. Conservation of mass tells us the number of atomic species on the left side of a reaction must equally what's on the right, it is literally a mathematical mass balance. Once we feed an artificial model a large dataset of known reactions and train the model accordingly, AI will be much better at predicting what's on the product side than humans. I personally see one of AI's biggest advantages is being able to develop correlations between large sets of research data.

We live in a data driven world, we just need help to better analyze the data we have.
I am an R&D chemist for a large pharma-medical company, I know how these things work from a clinical research side, I have a few questions based on your post.
Equations or chemical reactions is irrelevant to AI, all chemical reactions are essentially mathematical equations. Pretty much everything about chemistry is math, it's all an energy and mass balance equation. Conservation of mass tells us the number of atomic species on the left side of a reaction must equally what's on the right, it is literally a mathematical mass balance
So AI won't help researchers detect issues with biochemical reactions, which, for all we know was the failure of the Frequency Therapeutics' drugs as the reaction was unequal in lab rodents compared to humans? What's the point of it then? All it'll do is analyse the data and go "yep... you didn't fix tinnitus."
Once we feed an artificial model a large dataset of known reactions and train the model accordingly, AI will be much better at predicting what's on the product side than humans
And where does the dataset come from? How unbelievably large the data set will be if the AI doesn't "understand" chemical reactions and instead just references a data base.

The issue is there is very limited research into tinnitus and some people on here think that AI will propel the chance of success, all we need to do is 'train a model'. How do you feed an non-bias dataset into an AI model and have it account for the infinite number of variables beyond that which you have input? The point is you're developing a new drug to treat tinnitus, there is no large dataset of known reactions for this. There is no identified reaction within the brain that turns tinnitus off that an AI model can help translate to a treatment.
I personally see one of AI's biggest advantages is being able to develop correlations between large sets of research data.
This I do agree with. AI will be useful as a statistical analysis tool, but we're (at the very least) decades off it being able to solve complex research issues that our brightest and best scientists and engineers cannot.

I'm very much excited for the progression of AI, but people need to be realistic about it. Would love to be proven wrong and there's a lot of hype around AI right now, but expectations seriously need to be managed.

Anyway, the Frequency Therapeutics drugs are dead and I won't respond to any more AI questions now unless there's a breakthrough researching tinnitus with it.
 
Oh, thank god.
For someone that doesn't understand what execute mathematical equations (hint: to carry them out) means and presented a study where the authors say ChatGPT didn't provide the right answers and nothing more than little sniping comments, you sure are cocky. Keep living in your little ignorant bubble.

Username checks out.
 
For someone that doesn't understand what execute mathematical equations (hint: to carry them out) means and presented a study where the authors say ChatGPT didn't provide the right answers and nothing more than little sniping comments, you sure are cocky. Keep living in your little ignorant bubble.

Username checks out.
I mean you confuse LLMs with ML clustering problems, you clearly have no idea about what solving a measure theory problem entails (and you cherry picked the snippet you liked more) and you insist in confusing execution (hint: it means nothing) with solution... but for sure we can all learn one thing from you, how to be a proud paladin of the doom brigade. Don't bother replying with other unsubstantiated nonsense, you are ignored from now on.
 
Korro Bio and Frequency Therapeutics Announce Merger Agreement

Merger to create a Nasdaq-listed genetic medicines company focused on advancing Korro Bio's wholly owned portfolio of RNA editing programs

Lead program is a disease modifying therapy for patients with alpha-1 antitrypsin deficiency (AATD), with preclinical data showing an increase of normal A1AT protein to 85% of total protein in circulation

Combined company is expected to have cash balance of approximately $170 million at close, which is expected to provide cash runway through several value-creating milestones and into 2026

Companies to host conference call today at 8:30 a.m. ET
 
Merger to create a Nasdaq-listed genetic medicines company focused on advancing Korro Bio's wholly owned portfolio of RNA editing programs

Lead program is a disease modifying therapy for patients with alpha-1 antitrypsin deficiency (AATD), with preclinical data showing an increase of normal A1AT protein to 85% of total protein in circulation

Combined company is expected to have cash balance of approximately $170 million at close, which is expected to provide cash runway through several value-creating milestones and into 2026

Companies to host conference call today at 8:30 a.m. ET
@fletchermunson, so what?
 

Log in or register to get the full forum benefits!

Register

Register on Tinnitus Talk for free!

Register Now