Writing Frank Rosenblatt Back Into AI History

Writing Frank Rosenblatt Back Into AI History

Share

March 24, 2026
Amplify
Elevate
Stories
Frank Rosenblatt in his kitchen

Frank Rosenblatt at home in Chateau Rosenblatt; image courtesy of William Mutch

Dartmouth Faculty Revive a Machine Learning Pioneer’s Voice

Lunchtime author talks were introduced into the Libraries events cycle almost two years ago to amplify Dartmouth scholarship. Orchestrated by Rachel Starr GR ’13, these one-hour talks welcome everyone to hear from writers as they immerse us in their written world. Publications range from research findings and creative nonfiction to biographies, poetry, and young adult fiction. In their talks, Dartmouth authors illuminate an aspect of our social, economic, cultural, and political milieu. We may not walk out experts on the subjects shared with us, but we’re all the richer and more informed thanks to them. 

Like Dartmouth’s liberal arts mission and championing of interdisciplinary research, teaching, and learning, so many of the books discussed in these lunchtime author talks weave together multiple disciplines, demonstrating just how interconnected we and our world are. It’s true we don’t know what we don’t know. However, when we hear and witness new research, stories, and information, some of those unknowns come to light. It’s up to us to use that spark of new knowing as a force of curiosity and inspired action.

A Genre-Bending Work of Nonfiction

For James E. Dobson and Rena J. Mosteirin, professors who presented their print and open access published book, Perceptron, at a Libraries author talk, what they didn’t know galvanized them into action. During the pandemic, they saw an opportunity to lean into their shared profound curiosity about a long-dead, seemingly forgotten man, Frank Rosenblatt. What manifested was Perceptron, a genre-bending work of nonfiction blending poetics, art and community, science, warfare, politics, and biography.

Any history of modern AI [that excludes Frank Rosenblatt] is not just faulty, it’s impoverished.

Associate Professor James E. Dobson, Department of English and Creative Writing

Informed by extensive archival research and in-depth interviews with those who knew and loved Frank, Perceptron tells Frank’s personal story through imagined possibilities, letters and postcards, previously unavailable photographs, a love language, and invented poetic forms. It also colors in the space that time erased, not just of who Frank was as an everyday “renaissance man” but also as a scientist, friend and lover, and passionate researcher.

Frank Rosenblatt

image of Frank Rosenblatt from Wikimedia Commons

Technology Ahead of Its Time

Frank’s personal history is also machine learning’s origin story and thus also the origin of today’s artificial intelligence. While no one person can be given full credit for inventing AI, Frank Rosenblatt was part of a 1950s neural network, machine learning, and computer vision research zeitgeist. He was arguably a vanguard on the edge of something truly phenomenal. “Frank was briefly celebrated as an instigator in exciting new research but disappeared from public consciousness, and as such little is known about the short and brilliant life of one of the earlier [AI] developers,” Dobson mentioned at the author talk. Like so many who conduct research and then publish their results, not all efforts come with promising results or happy endings. 

The challenges Frank faced in his perceptron model research and testing in the 1950s mirror the same challenges that exist today. “The problems that people wanted the perceptron to solve, we want the large language models (LLMs) to solve,” says Dobson, “because we imagine them as a universal machine. The perceptron couldn’t do all the things it was thought to do, and it's the same with LLMs. We want them to be that, but they’re not that. Maybe the answer lies in future technologies.” As many times as Frank discovered moments of the phenomenal, he failed. His rollercoaster of success impacted his funding and how others regarded him in his fields of research. Then, while on a recreational sailing trip, Frank died on his birthday. He was 43. 

The problems that people wanted the perceptron to solve, we want the large language models (LLMs) to solve because we imagine them as a universal machine. The perceptron couldn’t do all the things it was thought to do, and it's the same with LLMs.

Associate Professor James E. Dobson, Department of English and Creative Writing
Frank Rosenblatt

image of Frank Rosenblatt working on the “electronic profile analyzing computer,” a precursor to the perceptron | courtesy of Cornell University Division of Rare and Manuscript Collections

To understand what inspired Dobson and Mosteirin to co-author Perceptron and their view on publishing open access as a way to amplify and broaden access to Dartmouth scholarship, they shared their thoughts in an interview. At the crux of their curiosity was the question of why and how Frank was “ostensibly ‘erased’ in AI academic writing and AI advancements in the decades following,” says Dobson. 

Reading Perceptron, you learn the complicated path to how he was erased, but one point on page 61 stands out: “Rosenblatt’s name was unavoidably linked to a particular program or brand of artificial intelligence. This branch was rooted in neural networks and what was referred to as ‘connectionism’ in the 1980s and 1990s. Today these are recognized as the most likely line of research and product development related to machine learning. In the 1980s and 1990s, while these technologies were continuing to be developed and expanded and even finding productive applications, they were mostly regarded as a dead end.”

Rewriting Frank Rosenblatt Into History

It was during the research for a book on computer vision published in 2023, The Birth of Computer Vision, Dobson discovered Frank Rosenblatt. While The Birth of Computer Vision couldn’t contain all that Dobson wanted to write and share about Frank, Dobson knew he had to learn more about Frank and formally document his findings for future generations. For Mosteirin, her curiosity and desire to reanimate Frank was a reflexive, almost protective act of solidarity and “seeing.” Mosteirin explains that Frank “was easy to make fun of because he didn’t stick to one track [of academic rigor], and those other works often failed, making it easier for fellow academics to disregard him.” Mosteirin continues, “Frank was a man of many talents, upholding the ideals of liberal arts that we pursue at Dartmouth. He didn’t professionalize his students; he wanted them to learn for themselves and often pointed them in multiple directions” much like his personal research practice. Frank “pushed at the edges of what interested him” rather than limiting himself to a single field.

Frank’s story is almost the opposite of what college is like today—he exemplified what it meant to grow and explore; that sort of thing is almost unimaginable.

Rena J. Mosteirin, Lecturer in the Masters of Arts in Liberal Studies Program
book cover Perceptron by James E. Dobson and Rena Mosteirin

Giving Frank Eternal Life

Dobson and Mosteirin’s choice of publisher was intentional and twofold. Firstly, a print publication wasn’t enough. Their goal was to enable anyone anywhere in the world to freely access Perceptron and, as such, have perpetual access to Frank’s machine learning and computer vision research and to know who he was beyond his scientific outputs. Open access, for Dobson and Mosteirin, was the best way to ensure all that. The book, Mosteirin explains, “is a symbol of Frank: experimental, and as such, analogous to his imagination and his life.” Publishing open access reverses the seeming erasure and disregard of Frank’s academic contributions to the AI field and gives his legacy to the people. 

Dobson and Mosteirin felt it was obvious they would publish with Punctum Books, an “independent queer- and scholar-led, community-formed, and peer-reviewed Sparkly Diamond open-access (OA) publisher” for books that “are genre-queer and genre-bending and take experimental risks with the forms and styles of intellectual writing.” Exactly the space Dobson and Mosteirin needed to publish Frank’s legacy to most fully illuminate him, his academic contributions, and his loves and vast interests. While there are minimal acknowledgments of him across the historical records, their writing Frank’s legacy and publishing it with Punctum permanently affixes Frank to that record. 

I love the idea that anyone can get the full text [of Perceptron] as they want it for free, anywhere and anytime, and to use on syllabi.

Rena J. Mosteirin, Lecturer in the Masters of Arts in Liberal Studies Program

The second reason Punctum was the best publishing house for this book was, as Dobson points out, because this sort of biography isn’t precisely “academic in nature.” Dobson and Mosteirin’s desire to write a book that blended some traditional academic writing with a biography of a gay man and creative poetics fit with Punctum’s mission. The publishers encourage works that push into more experimental spaces and cross discipline, genre, and form boundaries. “The scholars who lead Punctum are super interested in queer history and minor figures from the past, so they are personally invested in these unheard and untold stories and support pushing back on erasure,” notes Mosteirin. 

image of Frank Rosenblatt, left, and Charles W. Wightman working on part of the first perceptron in December 1958

image of Frank Rosenblatt, left, and Charles W. Wightman working on part of the first perceptron in December 1958; courtesy of Cornell University Division of Rare and Manuscript Collections

Translating Frank’s Voice & Work for Future Generations

For Mosteirin, centering love in Frank’s story was essential, and poetry was a way to “gesture at who he was…I tried to bring him to life and show the person who loved and was loved; had dreams and hopes and desires. He underwent great transformations in how he thought about things.” With Frank’s personal voice missing from history, Dobson and Mosteirin read numerous letters and postcards sent to Frank and conducted interviews with people who knew him, discovering a new language that translated Frank’s voice and way of seeing the world to the reader, “restor[ing] the person at the center.” Mosteirin discovered “moments of beauty and joy,” a hallmark of what it was like to be in Frank’s orbit. “He had a rich life, and we could get to it by tracing the human network around Rosenblatt, which was rewarding,” Dobson adds.

Archives are inspirational and encourage you to give things a go, to discover something amazing. Without archives, we can’t dig deeper into our own histories to better understand where we are going today and what our future could look like.

Rena J. Mosteirin, Lecturer in the Masters of Arts in Liberal Studies Program

Hearing Dobson and Mosteirin discuss their research process at the author talk, the audience gained a sense of how much care and attention they put into documenting Frank Rosenblatt's life and research. For Dobson, he hopes that readers today and in the future appreciate just how many projects Frank was involved in and how his approach to experimentation is a “gold mine” for future research. Frank was “incredibly connected to important researchers, technologies, social movements, and political events,” so returning to the archives is “incredibly generative. There’s always more to look at and think about.” For Mosteirin, what she hopes people learn from Frank, beyond his groundbreaking machine learning and computer vision modeling, is “not to box yourself in and to keep dreaming and imagining and being weird.” 

Read Perceptron
Back to top