Mike Magee
Let me be the first to introduce you to Claude Elwood Shannon. If you have never heard of him, but consider your own knowledge and involvement, including on the interface of AI and medicine, don't be embarrassed. I taught for a semester of “AI and Medicine” in 2024 and was only recently introduced to “Claude”.
First, let our product Claude be different from Claude. The man died a quarter of a century ago, except those in the depths of the AI field have been forgotten – so far.
In recognition, Claude Elwood Shannon is often called the “father of information theory.” He graduated from the University of Michigan in 1936, majoring in electrical engineering and mathematics. At the age of 21, as a master's student at MIT, he wrote a master's thesis titled “Symbolic Analytical Relay and Switching Loops”, among which ones were “Birth Certificates of the Digital Revolution”, and received the Alfred Noble Prize in 1939 (No, no, no, no, no That Nobel Prize).
In the early days, none of these were particularly obvious. A University of Michigan biopic claims: “If you were looking for a world changer in your 1936 UM class, you probably wouldn't pick out Claude Shannon. A shy, sticky young man from Gaylord, Michigan, sometimes without a hardworking and hopeless air, sometimes a fun giggle, but without a prejudiced person. Tightly knotted, his hair neatly splits his premium photos.”
But this is one of the historical misunderstandings of his alma mater. “That unassuming senior would go on to take his place among the most influential Michigan alumni of all time—and among the towering scientific geniuses of the 20th century…It was Shannon who created the “bit,” the first objective measurement of the information content of any message—but that statement minimizes his contributions. It would be more accurate to say that Claude Shannon invented the modern concept of information. Scientific American called his groundbreaking 1948 paper, “Mathematical Theory of Communication”, “Juna Kata of the Information Age”.
Just 5 days ago, the Washington Post technical columnist Geoffrey Fowler-Claude, not a human, introduced me to “Claude”. His post titled “5 AI Robots Perform Our Hard Reading Tests. One is the Smartest – This is Not Chatgpt,” I caught my attention. As he explained: “We challenged AI helpers legal contracts, simplified medical research, accelerated novels and understood Trump’s speech.”
The results of the judging medical research test are Scripps Research Translatation Institute, Eric Topol. 115 questions about two scientific research papers were asked about five AI products: the three-year outcomes of acute post-sequelae of Covid-19 and the retinal optical coherence tomography characteristics associated with events and prevalent Parkinson's disease.
Not to bury the lead, Claude was decisively won, not only in science but also with four names I'm familiar with – Google's Gemini, Open AI's Chatgpt, Microsoft Copilot and Metaai. This made me a little embarrassed. Never heard of Claude’s products?
For the answer, let's review the AI history.
The 2023 New York Times title announced that the rapid development of generative AI is “exciting and terrifying” after tracking four years of progress. “What we see is a machine that knows how to reason, is proficient in all human languages and is able to sense and interact with the physical environment,” their technology columnist wrote.
Dr. Leonid Zhukov, director of the Global Institute of Artificial Intelligence at Boston Consulting Group (BCG), then believes that products such as Chatgpt-4 and Genesis (Google's AI competitors) have the potential to become the brains of independent agents, the brains of independent agents – not only acting in their environment, but also doing that over the next 3 years.
Openai co-founders Elon Musk and Sam Altman initially expressed concern in 2016 about not only mastering language machines, but also thinking and feeling in a superhuman way. The desire for security and regulatory oversight has brought them early. But this didn't last long. When Musk failed to try to gain most control over the now successful Openai, he jumped and later launched his own business called “Xai”.
Meanwhile, the open AI board launched a coup and paid to Sam Altman, claiming that he was no longer regulated, but was all involved in the “arms race” of AI seeking profits. Sam lasted for a few more days before Microsoft had $10 billion in hands. Meanwhile, Google engineers who obtained the original classification through algorithms in 2016 created Genesis, and the comprehensive weapon competition began, now including Facebook's Metaai and Super-Powers Goggles.
Artman later wrote an op-ed titled “The Age of Intelligence,” explains, “Technology has brought us from the Stone Age to the Agriculture Age and then to the Industrial Age. From here, the road to the (AI-enabled) intelligence era paved with computing, energy and human will.”
Claude was born the same year. Its parents are the sibling co-founders of 2021 Public Benefits Company, Human, Dario Amodeo and Daniela Amodeo. They were the vice president of research and vice president of security and policy at OpenAI until Sam Altman converted a nonprofit organization into a capped for-profit entity (with Microsoft's wing) that caused high levels of tension and distrust that those entities and those who did not trust believed that security and public places were damaged. After all, the whole idea is to let Openai “build Safeai and share the benefits with the world”.
In December 2020, Dario, Daniela and 14 other Openai researchers jumped into the boat. Their new board recognizes the dual task: seeking profits for shareholders as part of their fiduciary responsibilities while creating “transformative AI that helps people and society thrive” and “pursuing the safety and ethics of AI when needed, rather than creating profits.” Their attitude toward “helpful and harmless” AI assistants is based on “Constitutional AI” on the first anniversary of 2022. This creation of humanity (AI Constitution) examines useful boundaries in the image of Claude Elwood Shannon, Claude, Claude, AI and Soul.
They accepted a technology used to develop, called “learning from human feedback” (RLHF). Definition: RLHF = “The model has an open conversation with the human assistant, producing multiple responses for each input prompt. Helpful and/or harmlessover time, reward the model of the model. “This allows anthropomorphism to have an open conversation with the human assistant, generating multiple responses for each input prompt.
As the process evolves, they are able to train AI to rate it with consistency with established constitutions. AI can now be harmless and helpful to itself. The new process, the new process of learning from AI Feedback (RLAIF) is the automation judge of RLHF. Taking a step further, Dario revealed the Constitution, which strengthens human commitment to transparency and public service.
What can Claude do? It is possible to generate text, summary, search, code and higher precision at the same time because it does not depend on the Internet to search for content. Now, researchers are quickly training at Claude to “respond based on character traits… etc..”
In addition to winning the Washington Post liberal arts test (including law, medicine, literature, and politics), the Claude website (free access) made 100 million visits in March 2025, and its iOS app had 150,000 downloads in the first week of its May 2024 release. The $18.2 billion raised by the crowd (2nd offare) raised 150,000 visits. Name its major cloud and training partners (AWS). Google also costs $2 billion.
On June 5, 2025, Dario wrote an OP in the New York Times titled “Don't Get AI Companies Off the Hook.” In it, he actively advocates transparency, noting that “it’s about responding to extraordinary times in a wise and balanced way.” Claude Elwood Shannon can be seen in the shadows, smiling quietly.
Mike Magee, MD, is a medical historian and a regular contributor to THCB. He is the author of Code Blue: The Inside Medical Industry Complex of America. (Grove/2020)
