Happy Monday Morning, Readers. Let’s be relentless this week!
I’m currently traveling in Thailand, so this edition of BioWire will be short and sweet. Nevertheless, I have some exciting updates to share, both in biotech and personally!
OpenAI Unveils New AI Model with Advanced Reasoning Skills
New Imaging Technique Makes Live Mice Transparent
Paxlovid Remains A Potent Antiviral Tool in the Ongoing Fight Against COVID-19
Before we launch into this week’s edition of BioWire, I have some exciting news to share! Earlier this year, I was approached by
for an invited guest article on his Substack, AI Supremacy (124K+ subscribers). True to its name, AI Supremacy focuses on the intersection of artificial intelligence with various technology fields. Michael was interested in my insights on how AI will impact and shape the future of drug development. After a summer hiatus, the article has finally been published, and I am very pleased with how it turned out. The reception has been overwhelmingly positive. For those who haven’t seen it, I strongly encourage you to check it out (link below)—I think you’ll find it an enjoyable read.OpenAI Unveils New AI Model with Advanced Reasoning Skills
The 2020s will undoubtedly be remembered for the introduction of AI tools to consumers, particularly through OpenAI’s ChatGPT-3.5 in late 2022. GPT-3.5 was a significant milestone as it was the first AI model to offer an interface accessible to the general public. This development reduced barriers to entry for regular consumers, allowing individuals without specialized knowledge to experience the power of advanced language processing capabilities. This breakthrough kicked off a race among many companies developing large language models (LLMs).
These LLMs have demonstrated impressive feats, yet they have also been prone to making rather silly mistakes. A well-known example circulating on the internet is how ChatGPT-4 struggles with a simple question: How many times does the letter ‘r’ occur in the word “strawberry”? Recognizing these shortcomings, the team at OpenAI took it as a challenge to enhance their model's reasoning abilities.
As of the last week, they have released their newest model ChatGPT o-1, which is claimed to have advanced reasoning skills over other classic models. So, does it fare any better with this question?
What's striking about o1's performance here is that it reflects genuine reasoning rather than mere data retrieval. The AI managed to solve this problem by analyzing each letter one by one in the word and counting the occurrences of the letter ‘r’. The hope is that the AI is truly solving problems outside of its training data, indicating true enhancement in cognitive capabilities. Projections based on current trends suggest that AI models could reach an IQ level of 140 by 2026.
For the biotechnology field, this rapid progress opens up exciting possibilities. Advanced AI could significantly accelerate research, assist in complex data analysis, and drive innovations that were previously unattainable due to computational limitations. As AI models continue to evolve, they may soon play a pivotal role in transforming biotech research and applications.
New Imaging Technique Makes Live Mice See-through
We owe a lot to the laboratory mouse, which has been a workhorse for the biomedical research community for at least 100 years. However, one of the main challenges with experiments in living systems is understanding what is happening inside. This often requires either crude, low-resolution imaging techniques or euthanizing the animal to view thin tissue sections under a microscope. This imaging challenge is largely due to light scattering caused by the complex structures inside organisms. Traditional optical imaging methods either don't penetrate deeply enough or aren’t suitable for live animals. However, a recent breakthrough offers a counterintuitive solution: using common food dyes to make live animal tissues temporarily transparent.
Researchers have discovered that introducing strongly absorbing molecules, specifically dyes like tartrazine (a yellow food color approved by the FDA), can reduce light scattering in tissues (Ou et al, 2024). The science behind this lies in the Lorentz oscillator model and the Kramers-Kronig relations. These principles predict that dyes absorbing light in the near-ultraviolet and blue regions can increase the refractive index at longer wavelengths without additional absorption. By matching the refractive indices between water and lipids in tissues, the dye effectively reduces scattering, rendering the tissue transparent.
This innovative approach was demonstrated in live rodents, where applying tartrazine made the skin, muscles, and connective tissues temporarily see-through. This transparency allowed for high-resolution imaging of structures deep within the body, such as fluorescently labeled neurons in the gut and cerebral blood vessels, without invasive procedures. While some limitations remain—like incomplete scattering reduction and dependence on dye diffusion—the method opens up exciting possibilities for non-invasive imaging in live animals, potentially transforming biomedical research and diagnostics.
Paxlovid Remains A Potent Antiviral Tool in the Ongoing Fight Against COVID-19
As COVID-19 continues to evolve, so does the understanding of treatments like Paxlovid (nirmatrelvir-ritonavir), the only FDA-approved oral antiviral therapy for mild to moderate cases in high-risk individuals. Recent observational studies reaffirm Paxlovid's effectiveness in reducing hospitalizations and deaths, even among vaccinated populations and those infected with newer Omicron subvariants (Rubin et al, 2024). This is significant because the original clinical trials focused on unvaccinated patients infected with the Delta variant, leaving questions about its current efficacy.
Beyond treating acute infections, scientists are exploring Paxlovid's potential in preventing and treating long COVID (PASC). While some studies suggest that early administration of Paxlovid may reduce the risk of developing long COVID symptoms, results are mixed, and more research is needed. Clinical trials are underway to assess its efficacy in patients already suffering from long COVID, aiming to uncover new therapeutic avenues. Additionally, Pfizer is developing a second-generation antiviral that doesn't require ritonavir, aiming to reduce drug interactions and side effects, which could enhance patient compliance and treatment outcomes.
Support:
These newsletters take a significant amount of effort to put together and are totally for the benefit of the reader. If you find these explorations valuable, there are multiple ways to show your support:
Engage: Like or comment on posts to join the conversation.
Subscribe: Never miss an update by subscribing to the Substack.
Share: Help spread the word by sharing posts with friends directly or on social media.
References:
Ou, Z., Duh, Y.S., Rommelfanger, N.J., Keck, C.H., Jiang, S., Brinson Jr, K., Zhao, S., Schmidt, E.L., Wu, X., Yang, F. and Cai, B., 2024. Achieving optical transparency in live animals with absorbing molecules. Science, 385(6713), p.eadm6869.
Rubin, R., The Latest Research About Paxlovid: Effectiveness, Access, and Possible Long COVID Benefits. JAMA.
Great article! Makes me excited about the next few years. Also, very cool techno-strawberry.
Colleague at work was marvelling at the tartrazine result the other day. It's really impressive.
Congrats on the AI Supremacy post, too. Nice work, David!