• AI KATANA
  • Posts
  • How artificial intelligence can revolutionise science

How artificial intelligence can revolutionise science

Also: Arm’s IPO lures retail traders seeking exposure to AI

Author

AI KATANA
September 17, 2023

Welcome!

In today's newsletter, we delve into the transformative potential of AI in scientific research, heralding what some predict to be a new era of unparalleled discovery akin to the advent of the microscope or research labs. Emory University's research suggests that ChatGPT may provide more accurate online medical advice than some existing platforms, at least in the domain of ophthalmology. In the financial realm, ARM's IPO captivates retail traders, reflecting their bullish outlook on AI technologies. SoftBank, riding the wave of its successful Arm IPO, eyes strategic investments in the AI sector, with OpenAI being a significant potential partner. In tech news, PhotoRoom teams up with Google Cloud to leverage the A3 supercomputer, promising a quantum leap in image processing speeds and capabilities. Lastly, on the venture capital front, European startups lag considerably behind their US counterparts in securing generative AI funding, with OpenAI's massive intake highlighting the disparity.

Sliced:

  • 🧬 How artificial intelligence can revolutionise science

  • 🏥 ChatGPT may be more accurate than other online medical advice

  • 📈 Arm’s IPO lures retail traders seeking exposure to AI

  • 💵 SoftBank seeks OpenAI tie-up as Son plans deal spree after Arm IPO

While concerns about the potential negative impacts of AI often dominate the discourse, there is growing optimism about its capability to catalyze major scientific breakthroughs. Historically, innovative tools and approaches, such as the microscope in the 17th century or the establishment of research laboratories in the 19th century, have been key drivers of scientific advancement. Today, AI's growing integration into various scientific domains suggests a new paradigm shift. Its capabilities range from analyzing vast amounts of data to modeling intricate systems, offering revolutionary insights in fields from drug discovery to astrophysics. Two groundbreaking applications of AI include "literature-based discovery" which sifts through existing literature to identify overlooked connections, and "robot scientists" that autonomously generate and test new hypotheses. Despite the promise, the adoption of AI-driven methods hinges on the scientific community's readiness to embrace these technologies, highlighting the need for adequate training and open standards. Mirroring historical sentiments, the incorporation of AI into science could usher in an era of unparalleled discovery and innovation.

Researchers from Emory University School of Medicine conducted a study evaluating ChatGPT's proficiency in diagnosing eye-related complaints. The results, published on medRxiv, revealed that ChatGPT compared favorably with human doctors in diagnosing symptoms based on the same set of data and outperformed WebMD's symptom checker. Despite the known "hallucination" issues with ChatGPT, in this study, it did not make any "grossly inaccurate" statements. Riley Lyons, an ophthalmology resident at Emory, and Nieraj Jain, an assistant professor, both expressed surprise at ChatGPT's relative proficiency. However, while ChatGPT's accuracy could potentially address some health care gaps, concerns remain about integrating AI into health care without the safeguards applied to other medical tools.

ARM's recent IPO attracted significant attention, especially from retail traders eager to invest in AI-related ventures. The chip designer witnessed over 20,000 buy orders on Fidelity's trading platform, outperforming popular stocks like Tesla and Nvidia. Starting with an IPO price of US$51, the stock opened at US$56.10 and closed 25% higher. Despite Fidelity advising investors to ensure their investments align with personal objectives, the high demand on platforms like Stocktwits indicates ARM's popularity among individual investors. The enthusiasm for ARM's IPO reveals a robust appetite for fresh tech stocks, following a lull in the IPO market over the past two years. Upcoming IPOs, including Instacart and Klaviyo, are keenly awaited, with SoFi Technologies offering shares for Instacart's IPO to individual platform users. The performance of these IPOs will be scrutinized as other companies, such as VNG and Birkenstock, prepare to go public.

SoftBank, bolstered by the successful IPO of UK chip designer Arm, which raised almost $5bn, is exploring major investments in the artificial intelligence (AI) sector, including a potential tie-up with OpenAI. Masayoshi Son, the founder and CEO of SoftBank, is contemplating putting tens of billions into AI following the Arm listing. Among the potential partnerships is one with OpenAI, the entity behind ChatGPT and backed by Microsoft. SoftBank is also considering alternatives, including significant investments in OpenAI's direct competitors. It has also shown preliminary interest in acquiring Graphcore, a UK-based AI chipmaker. Analysts predict that SoftBank, after Arm's IPO, may have access to around $65bn for investments, leveraging its own funds and its remaining stake in Arm.

🛠️ AI tools updates

PhotoRoom has collaborated with Google Cloud to harness the power of Google's A3 supercomputer for scaling large AI models and enhancing image processing capabilities. This partnership boosts PhotoRoom's capacity to process about two billion images annually and introduces new generative AI features, including a trained diffusion model. CEO Matthieu Rouif emphasizes that this synergy will help manage a rising number of global images without sacrificing quality or performance, enabling PhotoRoom to generate AI images 10 times faster than competitors. Central to this venture are features like Remove Background, Instant Backgrounds, and Instant Shadows that significantly benefit businesses by expediting product shot creation and obviating the need for expensive photo studios, thereby offering cost-effectiveness and efficiency.

💵 Venture Capital updates

Since 2019, European startups have secured only $1bn of the total $22bn venture capital funding for generative AI, with a mere 5% share, based on Dealroom's data. Conversely, US companies, significantly influenced by the success of OpenAI, claimed a dominant 89% with $20bn. OpenAI alone attracted nearly $12bn following its groundbreaking release of ChatGPT, cementing its leadership in the GenAI sector. The Bay Area, housing OpenAI's headquarters, received over $18bn, marking it as the heart of generative AI innovation. Despite trailing behind US leaders, European cities like London, Berlin, Amsterdam, and Stockholm emerged as notable players in the GenAI landscape.

🫡 Meme of the day

⭐️ Generative AI image of the day