Why Physics Majors Can’t Get Hired (Easily) Kyle Kabasares 2026-09-25 | Why Physics Majors Can’t Get Hired (Easily)
AI Models (o1, DeepSeek, QwQ, Gemini, Claude) Solve a College-Level Astrophysics Problem Kyle Kabasares 2025-01-03 | I came up with an astrophysics problem at the junior/senior college undergraduate level and wanted to see if some of the state of the art AI models could solve this problem.The Astrophysics problem and its solution can be read here: https://woozy-caper-657.notion.site/Neutron-Star-Problem-for-AI-Models-170b741346f2809ba52ddbf3c8325910?pvs=4
When you finish “Introduction to Electrodynamics” by Griffiths and think you understand the subject Kyle Kabasares 2025-01-02 | ...
OpenAI o1 Plays Connect 4 Against Me | Final Stream of 2024! Kyle Kabasares 2024-12-31 | Last stream of the year! Tonight, I will take on OpenAI's o1 in a game of Connect 4 and see who wins.
Fellow #researchers and #scientists, are you concerned about having to review #AI-generated papers? Kyle Kabasares 2024-12-31 | ...
ChatGPT Advanced Voice + Live Vision Plays Chess (Against Me AND ITSELF) Kyle Kabasares 2024-12-29 | I test out the Live Vision capabilities of the ChatGPT Advanced Voice model through a game of chess. In the first game, I get it to play me, and in the second, I get it to play against itself!
OpenAI Advanced Voice, o1, and o1-Pro Play Chess Against Me | X-mas 2024 Live Stream | Timestamps Kyle Kabasares 2024-12-26 | Merry Christmas to all! On this Christmas night, I decide to test ChatGPT's chess-playing skills. Christmas chess is always a bit special for me since it was Christmas 2013 that I got my first ever chessboard and properly learned how to play the game. I don't consider myself to be that good, but nonetheless, I want to see how ChatGPT stacks up to me.Timestamps25:50 o1 blunders a bishop27:50 o1 suggest an illegal Queen move 29:00 o1 suggests an illegal Knight move 30:55 o1 suggests the same illegal Knight move 32:20 o1 suggests the same illegal Knight move (again) 35:35 o1 suggests the same illegal Queen move 37:26 Starting a new o1 chat window continuing the game from the last position37:45 New o1 suggests the same illegal queen move 49:30 o1 suggests moving a Knight that does not exist52:15 Normal play continues with the move d552:23 Kyle recognizes a potential checkmate threat against o1 54:10 o1 suggests a move that is not possible57:30 Kyle checkmates o1 58:00 o1 denies checkmate 1:08:00 game with o1-pro begins1:57:03 o1-pro blunders Rxd5
Can ChatGPT Advanced Voice Mode + Video Solve Physics Problems?(Inclined Planes + Circuits) Kyle Kabasares 2024-12-24 | To try everything Brilliant has to offer—free—for a full 30 days, visit brilliant.org/KyleKabasares/. You’ll also get 20% off an annual premium subscription.This video is sponsored by Brilliant.As part of their 12 Day shipping spree, OpenAI released video and screen sharing capabilities with their Advanced Video Mode. I wanted to see how well it could do on relatively basic physics problems you would see in introductory physics courses. Check out the results for yourself!
ChatGPTs Santa Mode Helps Me Wrap a Christmas Gift (Advanced Voice + Video Test) Kyle Kabasares 2024-12-23 | In the spirit of Christmas, I wrap a gift with the "help" of ChatGPT's Santa Mode with the Advanced Voice + Video feature. Maybe St. Nick wasn't the most helpful here, but his encouragement got me through to the end!
12/21/2024 Live Stream Re-Upload w/Timestamps: Sora, Veo 2, o3 ARC-AGI, Apollo Safety Research Kyle Kabasares 2024-12-22 | Re-uploaded footage of my live stream yesterday, slightly edited to get rid of some computer malfunction moments.Time Stamps of Key Moments:1:20 Google Veo 2 Video Watching8:16 Sora tomato generation13:50 Sora Spaghetti generation15:30 Sora clown on bike generation20:10 Sora black hole generation 22:40 Sora Andromeda galaxy generation25:00 Reading Francois Chollet's statement on OpenAI o3 40:20 Sora "Dancing Christmas Tree" generation43:40 Trying to "trick" o1-mini reasoning model by inserting irrelevant information into a physics problem55:00 Apollo Research on AI Deception
Live Discussion: 12 Days of OpenAI, o3 surpasses 75% on ARC-AGI, Gemini-Flash 2, etc Kyle Kabasares 2024-12-21 | These past 2 weeks have been crazy, and I'm happy to finally get an opportunity to talk with you all about it.
What I Do for NASA + the Bay Area Environmental Research Institute Kyle Kabasares 2024-12-20 | I discuss the poster I presented at the American Geophysical Union Conference 2024 held in Washington DC from December 9 - December 13 this year. The poster featured a project I've been working on since I joined Ames in October 2023.iPoster: agu24.ipostersessions.com/default.aspx?s=27-10-2D-90-28-44-C0-6B-26-7D-AD-11-00-0A-E7-34&guestview=trueWhat is USGS 3DEP? usgs.gov/3d-elevation-program/what-3depWhat is a DTM, DSM, and CHM: neonscience.org/resources/learning-hub/tutorials/chm-dsm-dtm#:~:text=The%20Canopy%20height%20model%20represents,the%20surface%20of%20the%20earth.NASA Ames Research Center: nasa.gov/ames
Discussing o1-Pros Performance on the 2024 Putnam Math Competition (LIVE) Kyle Kabasares 2024-12-16 | I gave OpenAI's o1-pro the 2024 Putnam Exam. A Google Doc that features the links to my chat are provided here: https://t.co/uzcoT5prns(From my X Post):Following @DanHendrycks, I also gave the @OpenAI o1-pro model the 2024 Putnam Math Exam. Based on my (non-expert) understanding of the questions and the solutions, it appears that the o1-pro model scored somewhere in the range of 80-90/120, which based on past Putnam exams would put it in the top 1-2% of all participants, though we still await the results for this year's competition. A Google Doc featuring links to my chats is here: docs.google.com/document/d/1dwtSqDBfcuVrkauFes0ALQpQjCyqa4hD0bPClSJovIs/edit?usp=sharingWhat impressed me even more was the time it took to produce its answers. Below is a breakdown of the "Think time" for o1-pro on each individual question:A1 (3m 47s)A2 (6m 25s)A3 (6m 18s)A4 (3m 41s)A5 (9m 59s)A6 (12m 1s) B1 (2m 41s)B2 (2m 31s)B3 (6m 49s)B4 (2m 13s)B5 (4m 37s)B6 (2m 19s)In total, it spent 1 hour, 3 minutes and 21 seconds thinking on this exam compared to 6 hours (3 hours across 2 days) that students would normally get. o1 pro's average time per problem was about 5 min, though there is some variance depending on the question. As I write this, I'm still not entirely sure how to feel. On one hand, I find myself impressed, but on the other, I feel almost as if...I was expecting this? Based on my experience with o1-mini and o1-preview, this performance from o1-pro doesn't feel all that surprising... and it makes me wonder if I'm becoming more desensitized to these AI models' capabilities. It's as if with each model progression, I have unreasonable expectations. Still, I have to admit that perhaps it won't be too long before those think times go down even more, maybe by a few more factors or even a whole order of magnitude, and its performance on Putnam 2025 and beyond will be undoubtedly No. 1 in the world compared to the thousands of math undergraduates. Only time will tell, but I think we're in for some big surprises in 2025.
Sora kinda sucks...but its HILARIOUS Kyle Kabasares 2024-12-15 | I show some of the video generations I created from my first time using Sora last night. Admittedly, they might not have been the most solid prompts for the model, but nevertheless, I encourage you to be your own judge.
Follow Up to Im Deleting Several LLM Math/Physics Testing Videos Today Kyle Kabasares 2024-12-15 | Following up from part 1: youtube.com/watch?v=973VVBE2iR0TLDR: Someone I know and respect is the one who reached out to me to take down the content, and I don't want to fight it out legally even if I have fair use arguments on my side. Will just pivot and take a more creative approach to testing these LLM models' physics and math abilities.
Im Deleting Several LLM Math/Physics Testing Videos Today Kyle Kabasares 2024-12-14 | As the title suggests, I'll be purging quite a bit of content from my channel involving copyright issues. Made mistakes by not consulting authors/publishers beforehand, and I do apologize for that. Will try my best to come up with novel and original problems if I continue testing in this manner.
Reacting to OpenAIs Sora Release and Googles Quantum Computing Breakthrough (12/9/2024) Kyle Kabasares 2024-12-10 | Today was quite the day in terms of releases. OpenAI finally released the long-awaited video generation tool, Sora, and Google made a breakthrough in quantum computing. Wild times we live in!Google Quantum Computing Paper: arxiv.org/pdf/2408.13687Google Blog Post: https://blog.google/technology/research/google-willow-quantum-chip/My past video on quantum computing + AI: youtube.com/watch?v=OdqXQj-xusE
PSA online angry men: Women don’t owe you anything. Thank you for coming to my TED talk. Kyle Kabasares 2024-12-03 | ...
Could DeepSeek R1-Lite-Preview, o1-Preview, Claude Sonnet 3.5, or Gemini 1121 Work at Google? (LIVE) Kyle Kabasares 2024-11-24 | I give some premier LLMs questions and variations of questions from the book, "Are You Smart Enough to Work At Google?"Some interesting findings from this session were that even slightly modifying the question's phrasing and the numbers used in each question could be enough to throw the models off the correct path. I hope you find this kind of testing interesting!
Zettili’s quantum mechanics textbook is the #goat #physics #quantumphysics Kyle Kabasares 2024-11-23 | ...
Gemini Experimental 1121 Did ~10 Weeks of Quantum Mechanics Research in ~10 Minutes Kyle Kabasares 2024-11-22 | I gave Gemini Experimental 1121 the same research task I had as a new graduate student in 2017. Back then, I was supposed to work on this problem over 10 weeks during the summer. Gemini 1121 barely needed 10 minutes.
I Made DeepSeek-R1-Lite-Preview and Gemini Experimental 1114 Solve This Integral Kyle Kabasares 2024-11-21 | With the hype around DeepSeek and Gemini Experimental 1114, I decided to use them for the first time today. I brought back an old prompt that I used in this video: youtu.be/xqAAaXLJ_CY?si=CYx-eigI97WUprHQThe results were a bit surprising...
Artificial Intelligence + Quantum Computing = ??? Kyle Kabasares 2024-11-19 | To try everything Brilliant has to offer—free—for a full 30 days, visit brilliant.org/KyleKabasares/. You’ll also get 20% off an annual premium subscription.I share my thoughts on the book: "Convergence Artificial Intelligence and Quantum Computing - Social, Economic, and Policy Impacts"Book: wiley.com/en-ie/Convergence%3A+Artificial+Intelligence+and+Quantum+Computing%3A+Social%2C+Economic%2C+and+Policy+Impacts-p-9781394174119Bloomberg Quantum Computing Documentary: youtube.com/watch?v=1_gJp2uAjO0&t=75sThis video was sponsored by Brilliant.
Should I share my random late night thoughts more often? 😅 #AI #artificialintelligence #climate Kyle Kabasares 2024-11-15 | ...
I Used Perplexity AI Search for the First Time Kyle Kabasares 2024-11-09 | I show my first time using Perplexity AI, the pro version, on both the Desktop App and on Perplexity's website. Timestamps:1:50 - First Prompt: Papers related to my PhD 2:50 - Second Prompt: Challenges in my area of research4:30 - Third Prompt: Another research question6:00 - Voice Mode8:40 - Using Perplexity to do the integral of ln(x)9:18 - Fibonacci Sequence question12:00 - Fictional story about Einstein14:10 - Image generation 15:40 - Perplexity makes a mistake
Reflecting on OpenAI’s o1-Full, preview, and mini Models Kyle Kabasares 2024-11-08 | I share some thoughts I’ve had since getting a glimpse of OpenAI’s full o1-model.
11/7 marks the birthday of Marie Curie, the only one to win a Nobel Prize in Physics AND Chemistry. Kyle Kabasares 2024-11-08 | ...
Post #election2024 thought: Scientists, we have to start speaking truth to power more than ever. Kyle Kabasares 2024-11-06 | ...
OpenAI o1 FULL Was Accidentally Released Early?! Lets Test It! Kyle Kabasares 2024-11-02 | Looks like ChatGPT o1 was released early last night for a brief couple of hours. I was able to prompt it a few times before it was taken down. The original Tweet announcing the leak: https://x.com/apples_jimmy/status/1852665477606293976Putnam Archive: kskedlaya.org/putnam-archiveTimestamps:00:45 - Knowledge update request1:12 - EHT M87 black hole picture2:00 - Putnam 2023 Exam Question A12:53 - Putnam 2023 Exam Question A23:48 - Putnam Exam 2023 Exam Question A34:28 - o1 Referencing Putnam Exam 2019 in its database?6:35 - AI Safety/Alignment Question7:33 Final Thoughts
Elon Musk Says Grok Is Great at Math – Let’s Test It with Ahmed’s Integral! Kyle Kabasares 2024-11-02 | A new style of video I’m trying out. I love music and would love to incorporate it some more in my videos if possible!The original Tweet that sparked my curiosity of testing this out: https://x.com/elonmusk/status/1851276299547132050Ahmed's Integral The Maiden Solution: arxiv.org/abs/1411.5169Music Used:Provided to YouTube by YouTube Audio LibraryMoving In The Shadows · The SoundlingsMoving In The Shadows℗ YouTube Audio LibraryReleased on: 2024-10-30Habanera (by Bizet) by BizetCreative Commons — Attribution 3.0 Unported— CC BY 3.0https://creativecommons.org/licenses/...Music provided by FreeMusic109 / freemusic109
ChatGPT Search: Use at Your Own Risk (A PhDs Perspective) Kyle Kabasares 2024-11-01 | I try using ChatGPT search to look for references related to my PhD dissertation topic. Long story short, a serious user needs to verify the truthfulness and accuracy of what it gives you. There were instances of incorrect sources as well as sources that flat out didn't exist. Timestamps:2:10 First search, with my own papers being cited!4:20 Incorrect paper citation6:05 Incorrect author attribution on quasi-real source7:33 Hallucinated paper that does not exist but is plausible sounding8:50 iPhone video lost, switch over to laptop recording11:15 Final thoughts and reflection
Can OpenAIs o1-mini Decipher my Secret Messages? Kyle Kabasares 2024-10-31 | I test o1-mini's ability to break relatively easy ciphers that I "cleverly" came up with at 7 AM. Is it impressive? Maybe only a little bit right now, but in the future, is this a problem we need to seriously consider? Watch and make your own judgement!Timestamps:1. 0:26 First Cipher (“Hi I am Kyle”) with replacement of letters to number order in alphabet.2. 1:35 o1-mini starts to decipher3. 2:30 Second Cipher (Cesar Cipher)4. 3:04 o1-mini attempts to decipher second message5. 3:54 Providing a hint 6. 4:37 Final message 7. 5:58 o1-mini’s final attempt (no hints)8. 6:55 Another hint9. 8:19 o1-preview attempt (no hints)10. 9:22 Post-video discussion on AI safety, codebreaking, and quantum computers
Which AI, ML, and Data Science Books Do I Recommend? Kyle Kabasares 2024-10-30 | I cover several books that I own that are on the topics of data science, machine learning, and artificial intelligence.Book List:1. Becoming a Data Head2. Think Like a Data Scientist3. Data Science from Scratch4. Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems (3rd Edition)5. The Data Science Handbook (2nd Edition)6. Ace the Data Science Interview7. Artificial Intelligence: A Modern Approach (4th Edition)8. Deep Learning with Python (2nd Edition)9. Deep Learning10. Understanding Deep Learning (udlbook.github.io/udlbook)
Anthropic’s Claude Computer Use CAN Play Chess…Badly Kyle Kabasares 2024-10-28 | I was able to get Anthropic's Claude Computer Use Demo to actually make moves on the lichess.org interface. However, it is not quite on par with Magnus Carlsen to say the least. Hope you find it entertaining nonetheless!
Saying Good-Bye to My Favorite Quantum Mechanics Textbook... Kyle Kabasares 2024-10-24 | I say an emotional good-bye to Zettili Quantum Mechanics 2nd edition...and say HELLO to Zettili Quantum Mechanics 3rd edition! I talk about what makes Zettili's QM book my absolute favorite all time in favor of the others featured in the video.Books Shown:Zettili's Quantum Mechanics: Concepts and Applications (3rd edition)Griffiths's An Introduction to Quantum Mechanics (3rd edition)The Feynman Lectures on Physics Vol. 3: Quantum Mechanics: https://www.feynmanlectures.caltech.edu/III_toc.htmlBaym's Lectures on Quantum MechanicsLandau and Lifshitz Non-Relativistic Quantum MechanicsSakura's Modern Quantum Mechanics (3rd edition)Carroll's Quanta and FieldsSusskind and Friedman's Theoretical Minimum: Quantum MechanicsMerzbacher: Quantum Mechanics (2nd edition)
Can Anthropics AI Agent Play Online Chess?! Kyle Kabasares 2024-10-23 | I read a comment on my previously uploaded video questioning whether or not Anthropic's AI Agent could play chess against both bots and real humans online. The results were unexpected to say the least.Check out the latest announcement: anthropic.com/news/3-5-models-and-computer-use
Using Anthropics New Agentic Computer Use for the First Time! Kyle Kabasares 2024-10-23 | Anthropic came out swinging today (10/22/2024) by releasing its new Claude Sonnet 3.5 and Claude Haiku 3.5. Additionally, they also released a new feature known as "Computer Use" which gives these models an Agent-like ability to control a computer. I tried the feature out for myself and documented my experience. Unfortunately, I failed to realize my microphone was muted before it was too late, so I could only provide post-recording commentary over the original footage. Apologies in advance!Anthropic's Announcement: anthropic.com/news/3-5-models-and-computer-use
Reacting to: How I Become a Homeless Physics Professor Kyle Kabasares 2024-10-21 | Many of you may have heard about Dr. Daniel McKeown‘s story about how he, as a lecturing professor at UCLA, could not earn enough to live near UCLA, went homeless, and is now facing disciplinary action from the university. However, many of you probably don’t know that Dan and I were at UC Irvine together and started our Physics PhDs in the same year. I share some of my thoughts on the news story and encourage you to follow Dan’s accounts to hear his story through his own words.Link to the News Story: ktla.com/news/ucla-professor-says-hes-homeless-due-to-low-payLink to Daniel's Channel: youtube.com/@dr.danielmckeownastrophysicsLink to UC Payroll: https://ucannualwage.ucop.edu/wage/Link to UCLA Endowment: uclafoundation.org/Finances
Googles Notebook LM Created a Podcast of My Physics PhD Thesis Kyle Kabasares 2024-10-21 | A few weeks ago, I streamed my first time using Google's Notebook LM. This is a platform designed to enhance a user's interaction with their documents. One of the coolest features about Notebook LM is its ability to create podcasts based on the documents you feed it. Well, I gave it my PhD dissertation to summarize and the result was interesting to say the least!Try Google's Notebook LM: https://notebooklm.google/My PhD Dissertation: escholarship.org/uc/item/4204805q
Live from Worldcoin’s “A New World”: Keynote Address with Alex Blania and Sam Altman (OpenAI CEO) Kyle Kabasares 2024-10-18 | The Keynote Address at the “A New World” event in San Francisco, CA. Features multiple people associated with the project such as Alex Blania and OpenAI CEO Sam Altman.
I Went to Sam Altman’s Worldcoin Event As a Non-Techie Kyle Kabasares 2024-10-18 | I decided to be brave and see what all the fuss was about world.org and their event "A New World" in my home city of San Francisco. As someone who has absolutely zero clue about how cryptocurrency works, I thought it would be an interesting experience. I was right. I learned a bit about WorldCoin (now World's) mission to help with human verification in the age of AI, and was introduced to their signature ORB. It kind of looks like that villain from the video game Portal. I chose not to have my eyes scanned at this event.Learn more about WorldCoinCNBC: youtube.com/watch?v=qACuAWJRFoAWorldCoin: youtube.com/watch?v=xgUSRBa0zKQSong: Joakim Karud - Waves (Vlog No Copyright Music)Music provided by Vlog No Copyright Music.Video Link: • Joakim Karud - Waves (Vlog No Copyrig...
Live from WorldCoin’s: “A New World” in San Francisco, CA (Pre-Keynote, Breakfast) Kyle Kabasares 2024-10-18 | A brief view around the venue of the “A New World” event held by WorldCoin in San Francisco.
The Time a Professor Said My Physics PhD Cohort Would Fail (Watch Before Entering a PhD Program) Kyle Kabasares 2024-10-17 | I talk about the time a professor at UC Irvine told my first-year Physics PhD wouldn't go on to have successful academic careers. It was a painful and shocking moment for many of us in that room and has stuck with many of us to this day. I highly encourage prospective graduate students to please watch the video and take what I say into consideration if you aim to become a tenured research faculty one day.Articles Mentioned in the Videoslate.com/human-interest/2015/02/university-hiring-if-you-didn-t-get-your-ph-d-at-an-elite-university-good-luck-finding-an-academic-job.htmlscience.org/doi/10.1126/sciadv.1400005
I Took an Official Mensa Intelligence Test | An Honest Conversation About Intelligence Kyle Kabasares 2024-10-15 | I took Mensa's qualification test to satisfy my own curiosity and see if I could get in. Truthfully, I don't care much for IQ tests and these high-IQ societies, but in the age of AI, I thought it might be a useful thing to do. I talk about what it all means (if anything) and how I think that it is generally a good thing that more people will have access to "high-IQ" AI models in the near term.Check out Tibees: youtu.be/h6e0QjWG55U?si=TrdmU8dGPeS-iAa6Check out Flammable Maths: youtu.be/tnowi3POpB0?si=rS-_Bwz9DPjJ0kzS
My Reaction to the 2024 Nobel Prize in Chemistry (as a non-chemist) Kyle Kabasares 2024-10-14 | To try everything Brilliant has to offer—free—for a full 30 days, visit brilliant.org/KyleKabasares/. You’ll also get 20% off an annual premium subscription.I react to the announcement that the Nobel Prize committee has awarded David Baker, Demis Hassabis, and John Jumper as the 2024 Nobel laureates in Chemistry.This video was sponsored by Brilliant.
ChatGPTs Advanced Voice Speaks in My Grandmothers Filipino Language (Bisaya/Cebuano) Again! Kyle Kabasares 2024-10-12 | My grandmother engages ChatGPT in yet another conversation in Bisaya. This time, my grandmother asks ChatGPT about some travel destinations in Türkiye, and even gets a bit personal with our AI companion...
The 2024 #NobelPrize in #Chemistry was announced today! This is my 60-second reaction. #NobelWeek Kyle Kabasares 2024-10-09 | ...
Addendum to the Nobel Prize in Physics video, I acknowledge Prof. Hopfield’s physics background! Kyle Kabasares 2024-10-09 | ...
Can OpenAIs o1-preview Ace the 2023 Putnam Exam? Kyle Kabasares 2024-10-08 | I put the o1-preview and o1-mini model's math abilities to the test by giving them the 2023 Putnam Math Exam, supposedly the hardest math test given every year. This test is outside the model's training data, since their knowledge cutoff is October 2023 and the test was released in December 2023.Putnam Archive: kskedlaya.org/putnam-archive*** POST STREAM ***o1-preview's final score: 49/120.Median Putnam exam taker's score: 10/120.This would have placed o1-preview just outside the top 100 of all test takers (over 4000).Take the final score with a grain of salt, I had o1-mini act as a grader and asked it to compare the official Putnam solution and grade it according to the rough Putnam scoring guidelines. Apparently one of my viewers tried the same thing and o1-preview would’ve scored higher in their case (top 50)
