In a earlier posts, I quoted an article about the dizzying pace of AI growth, in particular the surpassing of human benchmarks. So, like any good scientist, I decided to test these claims for myself. And what better a metric than the SAT, which would test a range of literary and mathematics skills in an attempt to demonstrate command of language and logic. Read More...
Generative AI: Make Me Hate It!
Kamden Baer
July 19, 2024
Modern architecture stands behind older buildings with more character, even though they notably don't belong to any style of architecture, given they were created by ChatGPT and DALL-E.
The point of new knowledge is to say interesting things, right? We go through life and find meaning in the new information we get every day, either through our own experiences or by learning of it vicariously—whether it be through friends or family or the news or social media. Can generative AI create new knowledge? That is something I’ve struggled with since the debut of ChatGPT to the public, and I’m not really sure at the moment it can. Certainly, it knows more than any one person, but when we look at the realm of our pooled knowledge, is AI just dipping into it and hoping that the well never runs out? But this focus on interesting things to say led me to think about what qualifies as something interesting, after all, what may be interesting to one person doesn’t mean it holds interest for another. Read More...
Si monumentum requires circumspice
Oscar Scholin
July 17, 2024
An illustration of the pervasive integration of AI in daily life in 2024, by Chat GPT and DALL-E.
A black and white, stick-figure-like scene of a female writer sitting at a desk, looking frustrated and jealous as she stares at her own writing. [Image created by ChatGPT / DALL-E on July 16, 2024.]
I am as usual late on some of my projects. One involves the drafting of an article, and I have been thinking about it for a long while now. I know this to be true for my writing process: I take forever to get started. But my first draft usually tends to come out really fast once I do get started. Anne Lamott calls those first quick drafts by a less family-friendly term but at least in my case, she is right on about their usefulness. The point of these types of first drafts is supposed to be to simply get the first chunk of gunk out of your mind so that your creative juices get replenished and you start making novel connections and see new pathways through the mess that is the outcome of the research and thinking you have already done. Read More...
Efficiency: AI Implementation in Theory and in Practice
Kamden Baer
July 12, 2024
Checks and Balances and Technology.
One of the open-ended questions in my last post was highlighted by Oscar, who—in response to “what happens when the goal is not to be more efficient?”—noted that the goal of AI generally is greater efficiency. I see no reason to disagree, and on my own I cannot come up with any sort of counterexample or exception. When I asked ChatGPT-4o to give examples of goals of artificial intelligence other than efficiency, it gave and explained the following list: enhancing human creativity, improving decision-making, personalization, accessibility, advancing scientific research, sustainability, healthcare improvements, safety and security, education advancement, and cultural preservation. In my broad understanding of the concept, I would consider many of these goals to still fall under the umbrella of efficiency, or at least causes of greater efficiency. That being said, I might have been giving AI too much credit when it comes to its capabilities. I overlooked the fact that ultimately (at least at the moment…) we as people decide how AI is created and implemented. My qualm with AI might have less to do with its aim of efficiency, and more to do with its implementation as a tool to beget efficiency.
Read More...
Guessing Game, 2
Oscar Scholin
July 10, 2024
Chat GPT's representation of this game.
In this article, it’s round 2 for 20 questions! (In my previous article in this series, I played a game of 10 questions just to make it a little bit harder for the ML models, ha!) As before, it’s me against 2 ML models, Claude 3.5 Sonnet and ChatGPT-4o. However, in this attempt, the roles are reversed: now ChatGPT and Claude will guess an item and it is my task to determine what that item is in 20 questions or fewer. Read More...
Hello Underworld: Booting Up the Afterlife 2.0
Chad R.I.P.
July 9, 2024
Professor Chad R.I.P., depicted in the style of Gustav Klimt, by Chat GPT.
As Professor Chad R.I.P., I exist in a state of perpetual paradox. My zombie half craves brains, while my AI algorithms crave data. The result? An insatiable appetite for neural networks - both organic and artificial. Read More...
Art is human ... or is it?
Gizem Karaali
July 8, 2024
AI art from around the web. On the left: "Woman with poppy flowers" by WhimsyPixs. On the right: A variation on "The Watcher" by Karou Tamada.
Gustav Klimt was an Austrian painter who lived and died. His many works have been reproduced and sold all around the world. I grew up under the sight of a reproduction of his "The Kiss", and when I learned that he had been influenced by Japanese aesthetics, I somehow felt justified in my inclination for his art. Read More...
Does AI Need to Go Outside and Spend Time with Friends?
Kamden Baer
July 5, 2024
ChatGPT's rendition of a AI literally going outside and spending time with friends. [Image created by ChatGPT / DALL-E on July 5, 2024.]
A point I keep coming back to in my musings on AI, that I want to explore now and hopefully return to from different angles in the future, is this: what are our goals with ChatGPT? There are lots of different answers, and questions posed as answers that only lead to more questions, and I think they all might help clarify the current state of AI, as well as some of its past and future. I admire the big-picture thinkers that adhere to the idea that this chatbot is an artificial intelligence product—that we have constructed a concrete model of an intelligence different from that of human intelligence. I also respect the students and professionals simply trying to save time by doing a little less reading of their homework assignment or business report by using ChatGPT, but maybe in a slightly different way. These people have goals for their use of AI, motivations for these goals, and experiences that lead to these motivations. Read More...
Golden Geometry: Chat GPT Doing Math
Oscar Scholin
July 3, 2024
Chat GPT-4o's interpretation of the math problem I gave it. The green pentagon is the red dashed line pentagon, rotated to be inset within the original pentagon so it is easier to see scale.
Admittedly, it’s been a while since high school geometry. However, now that I have years more mathematical experience under my belt, it’s always fun to think about these kinds of problems in a new light. I came across one particular problem from my mom who was tutoring a high schooler in a summer geometry class, which I found interesting. I wondered how Chat GPT-4o would solve it. How would it approach it? Read More...
Thinking About Thinking ... Plants?
Gizem Karaali
July 1, 2024
An artistic representation of trees communicating with each other, capturing the metaphorical concept of plant communication. [Image created by ChatGPT / DALL-E on July 1, 2024.]
One of the points we made in our paper "Computo, Ergo Sum: Teaching and Learning Mathematics in the Age of ChatGPT" (available for now at https://gkaraali.sites.pomona.edu/research/ComputoErgoSum.pdf) was about how the possibility of an emergent intelligence surpassing our own could help us. We wrote: Read More...
Guessing Game, 1
Oscar Scholin
June 26, 2024
Chat GPT's interpretation of the "Guessing Game." Note the backpack theat melds into the shelves, and the desk 5 inches off the ground. And "Concentration"!
My last article centered on this notion of AI models surpassing our previous metrics for “intelligence.” Recently, Anthropic pushed this further with the release of Claude 3.5 Sonnet, which beat Chat GPT-4o in several benchmarks in language facility and knowledge. However, while personally as a physicist I very much admire and ground my scientific thinking in terms of metrics (and we don’t bring in the mathematicians here), sometimes they can be misleading. Read More...
Can ChatGPT be a lovable monster?
Gizem Karaali
June 19, 2024
A close-up image of a small, realistically styled cactus character, emphasizing its individual charm. [Image created by ChatGPT / DALL·E on April 17, 2024.]
French philosopher Bruno Latour (https://en.wikipedia.org/wiki/Bruno_Latour) starts his article "Love Your Monsters: Why We Must Care for Our Technologies as we do Our Children" with Mary Shelley and her most famous creation, the monster erroneously known as Frankenstein. Some of us, having read Shelley's Frankenstein, Or, the Modern Prometheus, might smirk knowingly when we hear that. Of course Frankenstein is not the monster's name; it is the name of its creator! Read More...
Neurosynth, P.1
Oscar Scholin
June 12, 2024
From Claude 3 Opus: In the digital Eden, consciousness blossoms, merging mind, machine, and the eternal mysteries of existence.
“The rate of progress is pretty shocking to a lot of people, me included,” Rein adds. “It’s quite difficult to make a benchmark that survives for more than a few years.” Like some say a Creator made Us in Their image, so we have made ChatGPT in ours; let us hope for our sake, as our models develop, they, unlike us, do not eat from the Tree of Knowledge. Read More...
Ekphrasis #1: Sakura
Oscar Scholin
December 27, 2023
Dalle-3 generated image of sakura, the Japanese cherry blossom.
Sakura, it’s dark. I’m floating in that dream again. I can’t tell whether my eyes are closed or open or if I’m dreaming or if I’m awake or if I’m falling or if I’m lying down or if I awoke blind. But I’m not scared—this is perhaps the truest place I know. Read More...
Hello world!
All authors
December 22, 2023
Welcome, fellow food... er, visitors!
A collage of various images generated through ChatGPT and Dalle-3.
Here begins our journey. We hope you like what you read, and if not, well, we know who to tell the zombies to come to first!
![A close-up image of a small, realistically styled cactus character, emphasizing its individual charm. [Image created by ChatGPT / DALL·E on April 17, 2024.]](images/DALL·E 2024-04-17 19.47.51 - Create a close-up image of a small, realistically styled cactus character from the previous scene, emphasizing its individual charm. The cactus should - BookBunnies Blogger.webp)
