By MIKE MAGEE<\/p>\n
Let me be the first to introduce you to\u00a0Claude Elwood Shannon. If you have never heard of him but consider yourself informed and engaged, including at the interface of AI and Medicine, don\u2019t be embarrassed. I taught a semester of\u00a0\u201cAI and Medicine\u201d\u00a0<\/a>in 2024 and only recently was introduced to \u201cClaude.\u201d<\/p>\n Let\u2019s begin with the fact that the product, Claude, is not the same as the person, Claude. The person died a quarter century ago and except for those deep in the field of AI has largely been forgotten \u2013 until now.<\/p>\n Among those in the know, Claude Elwood Shannon is often referred to as the \u201cfather of information theory.\u201d He graduated from the University of Michigan in 1936 where he majored in electrical engineering and mathematics. At 21, as a Master\u2019s student at MIT, he wrote a Master\u2019s Thesis titled \u201cA Symbolic Analysis Relay and Switching Circuits\u201d which those in the know claim was\u00a0\u201cthe birth certificate of the digital revolution,\u201d<\/a>\u00a0earning him the\u00a0Alfred Noble Prize<\/a>\u00a0in 1939 (No, not that Nobel Prize<\/a><\/em>).<\/p>\n None of this was particularly obvious in those early years. A\u00a0University of Michigan<\/a>\u00a0biopic claims, \u201cIf you were looking for world changers\u00a0in the U-M class of 1936, you probably would not have singled out Claude Shannon. The shy, stick-thin young man from Gaylord, Michigan, had a studious air and, at times, a playful smirk\u2014but none of the obvious aspects of greatness. In the Michiganensian yearbook, Shannon is one more face in the crowd, his tie tightly knotted and his hair neatly parted for his senior photo.\u201d<\/p>\n But that was one of the historic misreads of all time,\u00a0according to his alma mater.<\/a>\u00a0\u201cThat unassuming senior would go on to take his place among the most influential Michigan alumni of all time\u2014and among the towering scientific geniuses of the 20th century\u2026It was Shannon who created the \u201cbit,\u201d the first objective measurement of the information content of any message\u2014but 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,\u00a0\u201cA Mathematical Theory of Communication,\u201d\u00a0<\/a>the\u00a0\u201cMagna Carta of the Information Age.\u201d<\/a><\/p>\n I was introduced to \u201cClaude\u201d just 5 days ago by Washington Post Technology Columnist,\u00a0Geoffrey Fowler<\/a>\u00a0\u2013 Claude the product, not the person. His article, titled\u00a0\u201c5 AI bots took our tough reading test. One was smartest \u2014 and it wasn\u2019t ChatGPT,\u201d<\/a>\u00a0caught my eye. As he explained, \u201cWe challenged AI helpers to decode legal contracts, simplify medical research, speed-read a novel and make sense of Trump speeches.\u201d<\/p>\n Judging the results of the medical research test was Scripps Research Translational Institute luminary,\u00a0Eric Topol.<\/a>\u00a0 The 5 AI products were asked 115 questions on the content of two scientific research papers :\u00a0Three-year outcomes of post-acute sequelae of COVID-19<\/a>\u00a0and\u00a0Retinal Optical Coherence Tomography Features Associated With Incident and Prevalent Parkinson Disease<\/a>.<\/p>\n Not to bury the lead, Claude \u2013 the product \u2013 won decisively, not only in science but also overall against four name brand competitors I was familiar with \u2013 Google\u2019s Gemini, Open AI\u2019s ChatGPT, Microsoft Copilot, and MetaAI. Which left me a bit embarrassed. How had I never heard of Claude the product?<\/p>\n For the answer, let\u2019s retrace a bit of AI history.<\/p>\n <\/span><\/p>\n The\u00a0New York Times headline<\/a>\u00a0in 2023 announced the rapid progress of generative AI\u00a0 as \u201cExciting and Scary\u201d after four years of tracking its\u2019 progress. Their technology columnist wrote, \u201cWhat we see emerging are machines that know how to reason, are adept at all human languages, and are able to perceive and interact with the physical environment.\u201d<\/p>\n Leonid Zhukov, Ph.D<\/a>, director of the Boston Consulting Group\u2019s (BCG) Global AI Institute,\u00a0believed\u00a0<\/a>then that offerings like ChatGPT-4 and Genesis (Google\u2019s AI competitor) \u201chave the potential to become the brains of autonomous agents\u2014which don\u2019t just sense but also act on their environment\u2014in the next 3 to 5 years. This could pave the way for fully automated workflows.\u201d<\/p>\n OpenAI co-founders\u00a0Elon Musk and Sam Altman<\/a>\u00a0in 2016 initially expressed concerns about machines that not only mastered language, but could also think and feel in super-human ways. Desires for safety and regulatory oversight linked them in those early years. But that didn\u2019t last for long. When Musk\u2019s attempts to gain majority control of the now successful OpenAI failed, he jumped ship and later launched his own venture called \u201cXAI.\u201d<\/p>\n In the meantime, the Open AI Board staged a coup, throwing Sam Altman over-board claiming he was no longer into regulation but rather all in on an AI profit-seeking \u201carms race.\u201d That only lasted a few days, before Microsoft, with $10 billion in hand, placed Sam back on the throne. In the meantime, Google engineers, who were credited with the original break- through algorithms in 2016, created Genesis, and the full blown arms race was on, now including Facebook with it\u2019s MetaAI with super-powered goggles.<\/p>\n Altman later penned an op-ed titled\u00a0\u201cThe Intelligence Age\u201d<\/a>\u00a0in which he explained, \u201cTechnology brought us from the Stone Age to the Agricultural Age and then to the Industrial Age. From here, the path to the (AI enabled) Intelligence Age is paved with compute, energy, and human will.\u201d<\/p>\n Claude was born\u00a0<\/a>that same year. Its\u2019 parents were sibling co-founders of the 2021 public-benefit corporation, Anthropic, Dario Amodeo and Daniela Amodeo. They were the VP of Research and the VP of Safety & Policy at OpenAI until Sam Altman\u2019s conversion of that non-profit into a capped for-profit entity (with Microsoft in the wings) created high levels of tension and distrust in upper ranks who felt safety and public-good had been compromised. The whole idea, after all, was for OpenAI to \u201cbuild safeAI and share the benefits with the world.\u201d<\/p>\n In December 2020<\/a>, Dario, Daniela and 14 other OpenAI researchers jumped ship. Their new Board endorsed a dual mission to: seek profit for shareholders as part of their fiduciary responsibility,\u201d while creating \u201ctransformative AI that helps people and society flourish\u201d and if need be \u201cpursue AI safety and ethics over creating profit.\u201d\u00a0Their approach to \u201chelpful and harmless\u201d AI assistants was anchored in a commitment to \u201cConstitutional \u00a0AI\u201d on their 1 year anniversary in 2022. This human creation (the AI Constitution) juries the boundaries of usefulness and safety.\u00a0In the image of Claude Elwood Shannon, Claude, the AI with a soul, was born.<\/p>\n They embraced a technique for development called\u00a0\u201cReinforcement Learning from Human Feedback\u201d (RLHF)<\/a>. Definition: RLHF = \u201cModels\u00a0engage<\/a>\u00a0in open-ended conversations with human assistants, generating multiple responses for each input prompt. The human then chooses the response they found most\u00a0helpful and\/or harmless<\/em>, rewarding the model for either trait over time.\u201d This allowed Anthropic to \u201cengage<\/a>\u00a0models in open-ended conversations with human assistants, generating multiple responses for each input prompt.<\/p>\n As the process evolved, they were able to train the AI to grade the AI on consistency to the Constitution they had established. The AI now was able to grade itself harmlessness and helpfulness. The new process,\u00a0Reinforcement Learning from AI Feedback (RLAIF)<\/a>\u00a0was now the automated judge of RLHF. Going one step further, Dario disclosed the Constitution which reinforced Anthropic\u2019s commitment to transparency and public service.<\/p>\n What can Claude do?\u00a0<\/a>In can generate text in tandem, summarize, search, code, and more with high accuracy since it does not rely on Internet search for content. Researchers are now fast at work training Claude to \u201cgenerate responses based on character traits \u2026like curiosity, open-mindedness, and thoughtfulness.\u201d<\/p>\n Besides winning the Washington Post liberal arts test (including law, medicine, literature and politics), the Claude website (with free access) had 100 million visits in March 2025, and its iOS app had 150,000 downloads within its first week of release in May, 2024. Anthropic has raised $18.2 billion as of May, 2025, (2nd only to OpenAI) with Amazon as its top investor at $8 billion in return for naming its cloud service (AWS) its primary cloud and training partner. Google is in as well at $2 billion.<\/p>\n On June 5, 2025, Dario penned an opEd in the New York Times titled\u00a0\u201cDon\u2019t Let A.I. Companies off the Hook.\u201d<\/a>\u00a0In it he argues aggressively for a focus on transparency stating \u201cThis is about responding in a wise and balanced way to extraordinary times.\u201d One can almost see Claude Elwood Shannon in the shadows, quietly smiling.<\/p>\n