By Mike Magee
Let me be the first to introduce Claude Elwood Shannon. If you have never heard of it, but you consult informed and committed, even in the AI and Medicine interface, it is not ashamed. I taught a semester of “Ia and Medicine” in 2024 and only recently introduced “Claude”.
Let’s start with the fact that the product, Claude, is not the same as the person, Claude. The person died a quarter of a century ago and, with the exception of the deep in the AI field, has been largely forgotten, so far.
Among those who know, Claude Elwood Shannon refers to the “father of information theory.” He graduated from the University of Michigan in 1936, where he specializes in electrical engineering and mathematics. At 21, as a master’s student at MIT, Hey a master’s thesis entitled “A relay of symbolic analysis and switching circuits” that those who know were “the birth certificate of the digital revolution”, which earned him the Alfred Noble award in 1939 That Nobel Prize).
None of this was particularly obvious in those early years. A biographical film from the University of Michigan states: “If you were looking for world changingmen in the 1936 UM class, you would probably have not pointed to Claude Shannon. The shy and thin young man of Gaylord, Michigan, had a scholar air and, sometimes, a playful smile, but none of the obvious aspects of the bravery in the bravery of Michiganens, Shannon, Shannon, is more flickering, is a stricter couple of shit, more than the beast in the bulge of Michiganens.
But that was one of the historical erroneous interpretations of all time, according to his Alma Mater. “That somessuming senior wouled go on to take his place among the must in influential michigan alumni of all time – and among the towering scientific geniuses of the 20th century … it was tannon who created the“ bit, ”The firststbestage – hasurement of the hectare of the hectare of the hectaremement of the hasurementemement or the hectarementmement of the hasurementemement or the Hatisment or the Hatism or the Hatism or Histación – Contributions.
They introduced me to “Claude” only 5 days ago by the Washington Post technology columnist, Geoffrey Fowler – Claude the product, not the person. His article, entitled “5 Bots of AI, took our difficult reading test. One was smarter, and it was Nart Chatgpt”, Caht My Eye. As he explained, “we challenge AI assistants to decode legal contracts, simplify medical research, read a novel and make sense of Trump’s speeches.”
Judging the results of the medical research test was scripps Research Translational Institute Luminary, Eric Topol. The 5 products were asked 115 questions about the content of two scientific research works: three-year results of after-year after-school sequelae to COVID-19 and retinal coherence tomography characteristics associated with incident diseases.
Not to bury the leadership, Claude, the product, decisively won, not only in science, but also in general against the competitors of the four -name brand with which it was familiar, Gemini of Google, Open Ai’s Chatgpt, Microsoft Copilot and Metaai. I white a little ashamed. How had he ever heard of Claude the product?
For the answer, let’s go back a bit of AI history.
The head of the New York Times in 2023 announced the rapid progress of the generative AI as “exciting and scary” after four years of tracking its progress. His technology columnist wrote: “What we see that arise are machines that know how to reason, they are experts in all human language and can perceive and interact with the physical environment.”
Leonid Zhukov, Ph.D, director of the Global AI Institute of Boston Consulting Groups (BCG), believing that they then offer offers such as Chatgpt-4 and Genesis (Google’s competitor) “have the potential to become the salmuelas of automatic Ont agents in the environment of the next 3 to 5 years.
The co -founders of Operai, Elon Musk and Sam Altman, in 2016, initially expressed Conerns about machines that not only dominated the language, but also could think and feel superhumanly. The wishes of safety and regulatory supervision linked them in those early years. But that did not last the lung. When Musk’s attempts to obtain the majority control of the now successful Openai failed, jumped and then launched his own company called “Xai”.
I looked, the open Board of AI organized a coup d’etat, throwing Sam Altman through an arometo claiming that he was no longer in regulation, but rather in an “arms race” of the profits of AI. That only lasted a few days, before Microsoft, with $ 10 billion in his hand, replaced Sam on the throne. On the average, Google engineers, who were attributed to the original rupture algorithms in 2016, created Genesis, and the complete arms race was activated, which now included Facebook with super powerful glasses.
Later, Altman wrote an opinion article entitled “The Era of Intelligence” in what he explained: “Technology took us from the stone age to the agricultural age and the industrial era. From here, the path to it (the enabled, the age of intelligence is”
Claude was born that same year. His parents were co-founders brothers from the Benefit Public-Benefit 2021, Anthrope, Dario A Amodeo and Daniela Amodeo corporation. They were the Vice President of Investigation and the Vice President of Security and Politics in OpenI until the conversion of Sam Altman of that non -profit organization in a profit with limits (with Microsoft in the wings) created high levels of tension and exhibition that felt safe. The idea, after all, was for Openi to “build Safai and share the benefits with the world.”
In December 2020, Dario, Daniela and another 14 Operai researchers jumped the ship. His new board supported a dual mission to: seek profits for shareholders as part of their fiduciary responsibility “, while creating” the transformative that helps people and society “and, if necessary,” pursues the security and ethics of AI on the creation of profits. “His approach to the” useful and harmless “assistants” was anchored in a commitment to “constitutional AI” in its 1 year of anniversary in 2022. This human creation (the constitution of AI) was born the limits of utility and security. Soul, soul.
They adopted a development technique called “Human feedback reinforcement learning” (RLHF). Definition: RLHF = “The models are involved in open conversations with human assistants, generating multiple answers for each input indicator. The human then chooses the answer he found the most Useful and/or harmlessRewarding the model for any trait over time. “This allowed the anthropics” to involve models in open conversations with human assistants, generating multiple answers for each entry applicant.
As the process evolved, they were able to train the AI to qualify the AI in consistency to the Constitution they had established. The AI was now able to qualify the harmless and help it. The new process, the learning reinforcement learning of AI (Rlaif) was now the automated RLHF judge. Going one step further, Dario revealed the Constitution that reinforced Anthrope’s commitment to transparency and public service.
What can Claude do? In can generate text together, summary, search, code and more with high precision, since it does not depend on the search for internet content. Researchers are now fast at work training Claude to “generate responses based on character features … such as curiosity, open mentality and consideration.”
In addition to winning the Washington Post liberal test (including law, medicine, literature and politics), the Claude website (with free access) had 100 million visits in March 2025, and its application of iOS had 150,000 downloads within its first week of launch in May 2024. Anthrope has raised $ 18.2 billion to May of May of May 2025 OpenI) with the Amazon at the Amazon in the Blue Balloo in the Astoro of $ 8. Appoint your cloud service (AWS) your main cloud and training partner. Google is also at $ 2 billion.
On June 5, 2025, Dario wrote an Oped in the New York Times entitled “Do not let the companies out of the hook.” In it aggressively argues an approach to transparency that states that “it is the response in a wise and balanced way towards extraordinary times.” You can almost see Claude Elwood Shannon in the shadows, smiling silently.
Mike Magee MD is a habitual medical historian and collaborator of THCB. He is the author of Code Blue: Inside America’s Medical Industrial Complex. (Grove/2020)
]