Author: Jason Cox

Posted on

The Next Word

“I’m just very curious—got to find out what makes things tick… all our people have this curiosity; it keeps us moving forward, exploring, experimenting, opening new doors.” – Walt Disney

One word at a time. It is like a stream of consciousness. Actions, objects, colors, feelings and sounds paint across the page like a slow moving brush. Each word adds to the crescendo of thought. Each phrase, a lattice of cognition. It assembles structure. It conveys scenes. It expresses logic, reason and reality in strokes of font and punctuation. It is the miracle of writing. Words strung together, one by one, single file, transcending and preserving time and thought.

I love writing. But it isn’t the letters on the page that excite me. It is the progression of thought. Think about this for a moment. How do you think? I suspect you use words. In fact, I bet you have been talking to yourself today. I promise, I won’t tell! Sure, you may imagine pictures or solve puzzles through spatial inference, but if you are like me, you think in words too. Those “words” are likely more than English. You probably use tokens, symbols and math expressions to think as well. If you know more than one language, you have probably discovered that there are some ways you can’t think in English and must use the other forms. You likely form ideas, solve problems and express yourself through a progression of those words and tokens.

Over the past few weekends I have been experimenting with large language models (LLMs) that I can configure, fine tune and run on consumer grade hardware. By that, I mean something that will run on an old Intel i5 system with a Nvidia GTX 1060 GPU. Yes, it is a dinosaur by today’s standards, but it is what I had handy. And, believe it or not, I got it to work! 

Before I explain what I discovered, I want to talk about these LLMs. I suspect you have all personally seen and experimented with ChatGPT, Bard, Claude or the many other LLM chatbots out there. They are amazing. You can have a conversation with them. They provide well-structured thought, information and advice. They can reason and solve simple puzzles. Researchers agree that they would probably even pass the Turing test. How are these things doing that?

LLMs are made up of neural nets. Once trained, they receive an input and provide an output. But they have only one job. They provide one word (or token) at a time. Not just any word, the “next word.” They are predictive language completers. When you provide a prompt as the input, the LLM’s neural network will determine the most probable next word it should produce. Isn’t that funny? They just guess the next word! Wow, how is that intelligent? Oh wait… guess what? That’s sort of what we do too! 

So how does this “next word guessing” produce anything intelligent? Well, it turns out, it’s all because of context. The LLM networks were trained using self-attention to focus on the most relevant context. The mechanics of how it works are too much for a Monday email, but if you want to read more see the paper, Attention Is All You Need which is key in how we got to the current surge in generative pre-trained transformer (GPT) technology. That approach was used to train these models on massive amounts of written text and code. Something interesting began to emerge. Hyper-dimensional attributes formed. LLMs began to understand logic, syntax and semantics. They began to be able to provide logical answers to prompts given to them, recursively completing them one word at a time to form an intelligent thought.

Back to my experiment… Once a language model is trained, the read-only model can be used to answer prompts, including questions or conversations. There are many open source versions out there on platforms like Huggingface. Companies like Microsoft, OpenAI, Meta and Google have built their own and sell or provide for free. I downloaded the free Llama 2 Chat model. It comes in 7, 13 and 70 billion parameter models. Parameters are essentially the variables that the model uses to make predictions to generate text. Generally, the higher the parameters, the more intelligent the model. Of course, the higher it is, the larger the memory and hardware footprint needed to run the model. For my case, I used the 7B model with the neural net weights quantized to 5-bits to further reduce the memory needs. I was trying to fit the entire model within the GPU’s VRAM. Sadly, it needed slightly over the 6GB I had. But I was able to split the neural network, loading 32 of the key neural network layers into the GPU and keeping the rest on the CPU. With that, I was able to achieve 14 tokens per second (a way to measure how fast the model generates words). Not bad!

I began to test the model. I love to test LLMs with a simple riddle*. You would probably not be surprised to know that many models tell me I haven’t given them enough information to answer the question. To be fair, some humans do to. But for my experiment, the model answered correctly: 

> Ram's mom has three children, Reshma, Raja and a third one. What is the name of the third child?

The third child's name is Ram.

I went on to have the model help me write some code to build a python flask based chatbot app. It makes mistakes, especially in code, but was extremely helpful in accelerating my project. It has become a valuable assistant for my weekend coding distractions. My next project is to provide a vector database to allow it to reference additional information and pull current data from external sources.

I said this before, but I do believe we are on the cusp of a technological transformation. These are incredible tools. As with many other technologies that have been introduced, it has the amazing potential to amplify our human ability. Not replacing humans, but expanding and strengthening us. I don’t know about you, but I’m excited to see where this goes!

Stay curious! Keep experimenting and learning new things. And by all means, keep writing. Keep thinking. It is what we do… on to the next word… one after the other… until we reach… the end.

Posted on

Time

Time conducting the swirl of the universe.

Do you have time?

Isn’t that a funny question? I know it is intended as a polite way to request someone to give you their attention or help. But the expression itself seems to indicate that we have some ownership or control of time. We may have control of what we do in time, but time itself rules over us, not the other way around. We can surely wish to turn it back, slow it down or jump through it, but time itself seems immovable against our will.

If there is a ruler of time, perhaps it is gravity. The theory of relativity tells us that gravity can bend time. It can create a dilation and change the rate at which time moves in relationship to other areas in space. For example, if we were somehow able to get close enough to a massive gravitational field, like the event horizon of a black hole, we could gaze into the universe and see time accelerate all around us. Millennia would pass by while only a second ticks by on our watch. Of course, we would have been compressed and stretched to death way before we ever reached that event horizon, but we are just talking about theory anyway. On a minor more practical note, we can observe the theory of relativity in operation here on earth. Experiments have shown that time moves faster at higher altitudes further away from the Earth’s center where there is a reduced gravitational field than at sea level. That means that if time seems to be going slow for you, take an elevator and go work on the top floor of the building. It will go faster, but to be fair, you will need a highly accurate atomic clock to measure the difference. Yes, this relativity stuff is fascinating and weird! But once again, even in those peculiar experiments, time rules.

Time is like an expert conductor. Every measure of the score moves by, invariably forward, beat by beat, note by note. It is an inescapable crucible. It proves and bakes the bread of our hope, our dreams and our plans. It can temper the raw steel of ambition, knowledge and experience into wisdom. It seeks no destination but never stops moving. Like an infinite canvas, it holds every beginning and every end. Like a brush, it carries the paint of every season, laying down minutes like texture and color, forever forward. Like a needle, it stiches our memories deep into the fabric of the past. Every moment. Every movement. Every minute. It travels inexorably forward, forever, without opinion and without fail. Time keeps moving.

Time is a gift. Life requires it and memories are made of it. Don’t waste it. Don’t lose it. Find it, savor it, and enjoy it!

We are at a new week in time. We have beats in front of us yet to be realized. We have memories to make and seconds to enjoy. Go make the most of it!

Have a great time!

Posted on

Hurriquake

The bands of the hurricane started arriving Saturday evening. Here in our usually sunny SoCal valley of Santa Clarita, the sky was painted with a cloud spangled gorgeous warm glow. There was something in the air. The pressure was on a steady decline. Blankets of blue and gray clouds stretched across the horizon, putting us to sleep with ominous warnings from the National Weather Service.

Sunday morning awoke with a breezy drizzle that became a downpour. There was a unique odor with this storm. It smelled of ozone, like when you accidentally let the smoke out of your circuit board (not that I’ve ever done that!). The subtle hints told of the origin of the storm coming from lightning infested waters. Here we were, experiencing a unique moment in our time in SoCal. A hurricane!

“Earthquake Detected, drop, cover, hold on!” Of course, our go-to natural disaster danger here in California is earthquakes, not hurricanes. Clearly, it was jealous of all the attention the hurricane was getting, so we got to experience a 5.1 magnitude quake in Ojai, CA. It was 50 miles from our house and enough to bounce us around in our home, swing the chandeliers and move pictures on the wall, all while we watched the rain pour buckets in our backyard. A double feature hurriquake weekend!

We measured 8.75 inches of rain! For some of us in SoCal, you probably spent some time “unflooding” areas by clearing out rarely used drains and gutters. I had cleared our side yard drain on Saturday but forgot the back yard gutter… It’s okay, it let me know. Once that attention was given, I watched the rain fall. The pools of water slowly rose over time and the drainage ways surged to life. Streets filled up. Gutters were roaring. Even the Santa Clara river was full! It was mesmerizing.

Monday is here. The storm has passed. Brilliant white clouds cover the sky. Some remaining power outages across SoCal are gradually coming back online. Each hour gets brighter. Life is stirring about again.

Breathe in, breathe out. There is something magical about the air after a storm. Go on a walk outside. The landscape is glistening with a refreshing scent. The emerald green lawns and shrubs seem to radiate life. Patches of sapphire blue are now floating across the sky.

I’m issuing a new alert… a new day is here! Soak it in. Enjoy every minute.

Posted on

The Unlimited Future

One step to the edge of impossible. And then, further.

“One step to the edge of impossible. And then, further.” – National Geographic

There has been a lot of excitement in the scientific community these last several weeks. First, there is the constant buzz about AI and the pending birth of a real-life artificial general intelligence like Marvel’s fictional J.A.R.V.I.S. (which is just a rather very intelligent system by the way). Then there is this incredible medical news about the experimental anti-cancer drug, Dostarlimab, which had an unprecedented 100% success rate in eliminating tumors. Imagine what that could do for our human family! And now, just this past week, we saw the excitement building over LK-99, a polycrystalline compound that was reported by a team from Korea University to be a room-temperature and ambient pressure superconductor.

The LK-99 news was particularly fascinating to me. And I’m not alone. The scientific community is buzzing about it and excitedly conducting experiments to replicate to confirm or disprove the discovery. One of the things they hope to observe is “flux pinning”. Have you ever heard of flux pinning? Well, I hadn’t, so I decided to check it out. It turns out that flux pinning is a characteristic of superconductors where magnetic flux lines are trapped in place within a material’s lattice structure (quantum vortices). This flux pinning locks the superconductive material within a magnetic field, causing it to levitate. Can you imagine whole worlds built of this material? It may look a lot like Pandora from Avatar! More importantly this leads to benefits like enhanced current-carrying capabilities, higher magnetic field tolerances, and reduced energy losses.

Implications are mind blowing! If a room temperature and ambient pressure superconductor can be fabricated, we could see things like massively reduced losses in power transmission, higher performing electromagnetic devices (e.g. MRIs, motors, generators), revolutionized transportation systems (e.g. maglev trains, lightweight and energy-efficient propulsion systems), faster low-power computing devices and of course, new insights into the fundamental nature of matter and the universe. Of course, LK-99 may not be the superconductor we are looking for, but the quest continues… and we are learning!

I love science! The systematic rigor, the tenacious pursuit of discovery, and the passionate pursuit of understanding our universe is who we are. We thirst for knowledge and hunger for new abilities. It motivates us. It propels us to adapt. It allows us to survive and thrive when conditions are threatening. It is our genius, and perhaps at times, our curse. We are restless and unsatisfied. But that insatiable curiosity compels us to discover, to explore, to test, to add to our knowledge, to create and become more than we were.

Look, I know I’m incurably optimistic to a fault. I know that there are disappointments and failures ahead of us as well. And to be fair, the path to the future can sometimes seem impossible. But oddly enough, it is at those moments that we discover something different and something new. We see, we learn, we step to the edge and we go further! The unlimited future awaits. Let’s go!

One step to the edge of impossible. And then, further.
Posted on

Dog Days of Summer

Summer! The signs are all around us. There are the sounds of neighborhood kids playing outside, crowds splashing at the community pool, and calendars bursting with vacation notices. Take a look and you will see bicyclist sailing along the paseos, beaches full of sun drenched visitors, families enjoying evening backyard barbecues, and of course, SoCal seeing daily excessive heat warnings. We are definitely in the dog days of summer. 

Do you know where that saying comes from, “the dog days of summer?” They are typically the hottest, most uncomfortable part of summer in the Northern Hemisphere where dogs sit around panting in the heat, but that isn’t the origin of the name. It turns out that the “Dog days” is an astrological event. It is so named due to the heliacal rising of the star system Sirius, the “Dog Star.” It’s the brightest star in the constellation Canis Major. The heliacal rising is a fancy way to say when a star first makes its appearance above the horizon at dawn, appearing along with the sun to herald a new day. The Greeks associated the appearance of Sirius with heat, drought and lethargy. I can relate! I’m ready for an afternoon nap and it’s still morning. 

Stay cool! Enjoy the dog days of summer as much as you can and get plenty of water if you go out. And if you get a chance, gaze up into the heavens sometime, follow Orion’s belt down to his best friend, the brightest star. See if you can spot Spot. He is up there, the dog in the sky welcoming us all to enjoy the dog days of summer.

References

Becky Little, “Here’s why we call this time of year the ‘dog days’ of summer,” National Geographic, July 16, 2021, accessed July 10, 2023, https://www.nationalgeographic.com/animals/article/150710-dog-days-summer-sirius-star-astronomy-weather-language.

Posted on

The Journey to AGI

Glowing singularity on a black background.

Every week, we hear announcements of new AI powered tools or advancements. Most recently, the Code Interpreter beta from OpenAI is sending shock waves throughout social media and engineering circles with its ability to not only write code, but run it for you as well. Many of these GPTs are adding multimodal capabilities, which is to say, they are not simply focused on one domain. Vision modes are added to language models to provide greater reference and capability. It’s getting hard to keep up!

With all this progress, it makes you wonder, how close are we to Artificial General Intelligence (AGI)? When will we see systems capable of understanding, learning, and applying knowledge across multiple domains at the same level as humans? It seems like we are already seeing systems that exhibit what appears to be cognitive abilities similar to ours, including reasoning, problem-solving, learning, generalizing, and adapting to new domains. They are not perfect and there are holes in their abilities, but we do see enough spark there to tell us that the journey to AGI is well underway.

When I think of AGI, I can’t help but compare that journey to our own human journey. How did each of us become so intelligent? Ok, that may sound presumptuous if not a bit arrogant. I mean to say, not in a brag, that all of us humans are intelligent beings. We process an enormous amount of sensory data, learn by interacting with our environment through experiments, reason through logic and deduction, adapt quickly to changes, and express our volition through communication, art and motion. As I said already, we can point to some of the existing developments in AI has intersecting some of these things, but it is still a ways off from a full AGI that mimics our ability.

Instincts

We come into this world with a sort of firmware (or wetware?) of capabilities that are essential for our survival. We call these instincts. They form the initial parameters that help us function and carry us through life. How did the DNA embed that training into our model? Perhaps the structure of neurons, layered together, formed synaptic values that gifted us these capabilities. Babies naturally know how to latch on to their mothers to feed. Instincts like our innate fear of snakes helped us safely navigate our deadly environment. Self preservation, revenge, tribal loyalty, greed and our urge to procreate are all defaults that are genetically hardwired into our code. They helped us survive, even if they are a challenge to us in other ways. This firmware isn’t just a human trait, we see DNA embedded behaviors expressed across the animal kingdom. Dogs, cats, squirrels, lizards and even worms have similar code built in to them that helps them survive as well.

Our instincts are not our intelligence. But our intelligence exists in concert with our instincts. Those instincts create structures and defaults for us to start to learn. We can push against our instincts and even override them. But they are there, nonetheless. Physical needs, like nutrition or self preservation can activate our instincts. Higher level brain functions allow us to make sense of these things, and even optimize our circumstances to fulfil them.

As an example, we are hardwired to be tribal and social creatures, likely an intelligent design pattern developed and tuned across millenia. We reason, plan, shape and experiment with social constructs to help fulfil that instinctual need for belonging. Over the generations, you can see how it would help us thrive in difficult conditions. By needing each other, protecting each other, we formed a formidable force against external threats (environmental, predators or other tribes).

What instincts would we impart to AGI? What firmware would we load to give it a base, a default structure to inform its behavior and survival?

Pain

Pain is a gift. It’s hard to imagine that, but it is. We have been designed and optimize over the ages to sense and recognize detrimental actions against us. Things that would cut, tear, burn, freeze and crush us send signals of “pain.” Our instinctual firmware tells us to avoid these things. It reminds us to take action against the cause and to treat the area of pain when it occurs.

Without pain, we wouldn’t survive. We would push ourselves beyond breaking. Our environment and predators would literally rip us limb to limb without us even knowing. Pain protects and provides boundaries. It signals and activates not only our firmware, but our higher cognitive functions. We reason, plan, create and operate to avoid and treat pain. It helps us navigate the world, survive and even thrive.

How do we impart pain to AGI? How can it know its boundaries? What consequences should it experience when it breaches boundaries it should not. To protect itself and others, it seems that it should know pain.

Emotions

Happiness, fear, anger, disgust, surprise and sadness. These emotions are more than human decorations, they are our core. They drive us. We express them, entertain them, avoid them, seek them and promote them. They motivate us and shape our view of the world. Life is worth living because we have feelings.

Can AGI have feelings? Should it have feelings? Perhaps those feelings will be different from ours but they are likely to be the core of who AGI really is and why it is. Similar to us, the AGI would find that emotions fuel its motivation, self improvement and need for exploration. Of course, those emotions can guide or misguide it. It seems like this is an area that will be key for AGIs to develop fully.

Physical Manipulation

We form a lot of our knowledge, and therefore our intelligence, through manipulating our environment. Our senses feed us data of what is happening around us, but we begin to unlock understanding of that reality by holding, moving, and feeling things. We learn causality by the reactions of our actions. As babies, we became physicist. We intuit gravity by dropping and throwing things. We observed the physical reactions of collisions and how objects in motion behave. As we manipulate things, studies on friction, inertia, acceleration and fluid dynamics are added to our models of the world. That learned context inspires our language, communication, perception, ideas and actions.

Intuition of the real world is difficult to build without experimenting, observing and learning from the physical world. Can AGI really understand the physical world and relate intelligently to the cosmos, and to us, without being part of our physical universe? It seems to me that to achieve full AGI, it must have a way to learn “hands on.” Perhaps that can be simulated. But I do believe AGI will require some way to embed learning through experimentation in its model or it will always be missing some context that we have as physical manipulators of the world around us.

Conclusion

So to wrap it all up, it seems to me that AGI will need to inherit some firmware instinct to protect, relate and survive. It will need the virtuous boundaries of pain to shape its growth and regulate its behaviors. Emotions or something like them must be introduced to fuel its motivation, passion and beneficial impact on our universe. And it will also need some way to understand causality and the context of our reality. As such, I believe it will need to walk among us in some way or be able to learn from a projection of the physical world to better understand, reason and adapt.

Fellow travelers, I’m convinced we are on a swift journey to AGI. It can be frightening and exciting. It has the potential of being a force multiplier for us as a species. It could be an amplifier of goodness and aide in our own development. Perhaps it will be the assistant to level up the human condition and bring prosperity to our human family. Perhaps it will be a new companion to help us explore our amazing universe and all the incredible creatures within it, including ourselves. Or perhaps it will just be a very smart tool and a whole lot of nothing. It’s too early to say. Still, I’m optimistic. I believe there is great potential here for something amazing. But we do need to be prudent. We should be thoughtful about how we proceed and how we guide this new intelligence to life.

Posted on

A Declaration of Happiness

Picture of 4th of July Party generated by AI Stable Diffusion.

“We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness.” – Declaration of Independence of the thirteen United States of America, July 4, 1776.

Bring out the flags, fireworks, hamburgers and watermelons! Tomorrow is Independence Day in the US. I love the phrase “pursuit of happiness. ” It has been bouncing around in my head all weekend. To be fair, I know that happiness can be fleeting. But it is worth pursuing. It requires practice, energy and determination. But the reward set before us is joy, satisfaction and a smile.

I often say that our goal as a team is to help our businesses create great compelling content, products and experiences, better, faster, safer and happier. Those qualifiers are important. That last one, “happier” is especially relevant. Working in entertainment, we ship happiness as a product, but it means more than that. It means creating a work environment that is as delightful and rewarding as the products we produce. I believe happy people produce happy products. In every job that must be done, there is an element of fun! If you don’t believe me, ask Mary Poppins.

It’s true! We can find happiness at work. Some of life’s greatest moments are at the other end of hard work and effort. There is a great happiness that can be found in a job well done and a mission accomplished. For those of us in technology, solving complex problems and engaging creative energy to design and produce software, systems and automation is both taxing and rewarding. The beautiful paradox is that those things that are the most difficult, challenging and fraught with anguish, are often the ones with the highest dose of happiness, delight and satisfying accomplishment.

How are we doing? Are we meeting our mission in making this a happy place to work? Last week I sent out an end of quarter “happiness survey” to my team (with a scale from 1, “not happy at all” to 5, “extremely happy”). The last several months have presented us with some difficult challenges. I very much wanted to hear from all of my team to see how we were doing. The results were encouraging and informative. They highlighted some of the good but also the challenges before us and areas we need to improve.

Candid feedback is pure gold! Leaders need clear signal. My team impressed me with their candor. I was delighted. I assembled my leadership team and we looked at every comment and will be taking action to follow up on every suggestion and concern raised. It is a priority for us, as it should be.

Now, for your part…. You matter. Your happiness matters. Life is what you choose to make it. What can you do to cultivate happiness for yourself? Today, I challenge you all to think about how you are pursuing happiness. What does it look like for you? What can you change to make it better? Find your path to happiness. Look for it in your work, pursue it with all your heart and enjoy it. I wish you happiness this week and always!

For all of you in the USA, Happy Independence Day!

4th of July Party Image with hamburger, watermelon, US flags and fireworks.

Top blog post image of “4th of July Party” generated by Stable Diffusion AI and second image by Dall-E AI.

Posted on

Storm Warning

The sky was on fire. Lightning danced across the dark horizon like a shimmering lake of eels. They seemed to be swimming towards us. It was beautiful! Trees began swaying. The tornado sirens across the city spun to life and began screaming their warning of approaching doom. In the distance, exploding power lines and erupting transformers illuminated row after row of trees bowing all the way to the ground as the event horizon of the storm made its way towards us. I was in awe.

The atmosphere was electric and full of an eerie green light that gave way to the howling wind. Its roar turned into chords. It reminded me of an approaching train at the Metrolink station. Leaves and branches began pouring into the house like a pack of dogs chasing a racoon. I desperately hung onto the door. Soon, the wind nearly pushed me to the ground. Then I heard a voice. “Dad, what are you doing?! Shut the door!”

I confess, I can be curious to a fault. Last week we were in Tulsa, Oklahoma, visiting family. News hit of a devastating thunderstorm with 100 mph wind coming our direction. I had to see it firsthand. So, of course I opened the door. Eventually after the wise admonition from my daughters, I managed to get it closed and we secured ourselves with all the necessities stored away in the safe room.

It was dark. The wind and sirens continued for another half hour as the front of the storm crashed against the house. It creaked and shuddered with the booming thunder and pounding rain. After a while, the brunt of the storm moved through and left us with a peaceful heavy rain. Power was out. The only light came from everyone’s phones and the strobing lightning outside. For the next three days, the city would be without power. Trees covered the streets. Power poles were bent and broken to bits like toothpicks. Cellular towers were down and the internet was offline.

What a vacation! My family generally doesn’t like to go camping, so camping came to us. Traffic lights were out and grocery and gas stations were closed. We had to go back to the basics. I panicked for a while, but then I figured out how to make cowboy coffee.

Life is unexpected, terrifying and wonderful. You never know what adventure awaits. Be ready for it. Dare to take a look! Be curious and listen to the voice of wisdom. Which in my case, was the voice of my daughters.

Posted on

JasonGPT-1 : Adventures in AI

Distorted sci-fi black and blue world.

“Imperfect things with a positive ingredient can become a positive difference.” – JasonGPT

I don’t know how you are wired, but for me, I become intoxicated with new technology. I have a compulsive need to learn all about it. I’m also a kinesthetic learner which means I need to be hands on. So into the code I go. My latest fixation is large language models (LLMs) and the underlying generative neural network (NN) transformers (GPTs) that power them. I confess, the last time I built a NN, we were trying to read George H.W. Bush’s lips. And no, that experiment didn’t work out too well for us… or for him! 

Do you want to know what I have discovered so far? Too bad. I thought I would take you along for the ride anyway. Seriously, if you are fed up with all the artificial intelligence news and additives, you can stop now and go about your week. I won’t mind. Otherwise, hang on, I’m going to take you on an Indiana Jones style adventure through GPT! Just don’t look into the eyes of the idol… that could be dangerous, very dangerous!

Where do we start? YouTube of course! I have a new nerd crush. His name is Andrej Karpathy. He is a Slovak-Canadian computer scientist who served as the director of artificial intelligence and Autopilot Vision at Tesla and currently works for OpenAI. He lectured at Standford University and has several good instructional lectures on YouTube. I first saw him at the Microsoft Build conference where he gave a keynote on ChatGPT but what blew me away was his talk, “Let’s build GPT: from scratch, in code, spelled out.” (YouTube link). It’s no joke. He builds a GPT model on the works of Shakespeare (1MB), from scratch. After spending nearly 2 hours with him, Google Colab and PyTorch, I was left with a headache and some cuts and bruises. But I also had an insatiable desire to learn more. I have a long way to go. 

The way I learn is to fork away from just repeating what an instructor says and start adding my own challenges. I had an idea. I have done a lot of writing (many of you are victims to that) and much of that is on my blog site. What if I built a GPT based solely on the corpus of all my writing? Does that sound narcissistic a bit to you too? Oh well, for the good of science, we go in! Cue the Indy music. I extracted the text (468k). It’s not much, but why not? 

By the way, if you are still with me, I’ll try to go faster. You won’t want to hear about how I wasted so much time trying to use AMD GPUs (their ROCm software sucks, traveler beware), switched to CPUs, Nvidia CUDA and eventually Apple Silicon MPS (Metal Performance Shaders built in to the M1). All the while, I was using my fork of the code I built with Andrej Karpathy (ok, not him directly, but while watching his video). I started off with the simple Bigram NN Language model. And it is “Bi-Gram” not “Big RAM” but I found that to be ironically comical in a dad joke sort of way. 

My JasonGPT bigram.py started learning. It ran for 50,000 iterations and took about 8 hours. It even produced an output of random musings. While there was quite a bit of nonsensical output, I was amazed at how well this small run did at learning words, basic sentence structure and even picked up on my style. Here are some samples from the output I found interesting, comical and sometimes, well, spot on:

  • It’s a lot of time… But I think we also need science.
  • What are your big ideas?
  • Set our management to the adjacent ground (GND) pin.
  • I have a task to Disneyland out that this day.
  • I love the fun and fanciful moments as kids get to dream into their favorite characters, embrace the identity of their heroes, wrap themselves up starfish back.
  • Bring on the “power” of his accidental detail.
  • Your character provided faith, all kindness and don’t care.
  • Grab a difference too.
  • After several days of emailing, texting and calling, I received a text message.
  • Curl has the ability to provide timing data for DNS lookup, it will easily show or avoided.
  • Imperfect things with a positive ingredient can become a positive difference, just get that time.
  • I also believe we should exploit the fusion power that shows up each day in our company’s data.
  • Have you found a vulnerability? Are you concerned about some missing measures or designs that should be modernized or addressed? If so, don’t wait, raise those issues. Speak up and act. You can make a difference.
  • “I know what you are thinking.” the irony
  • We are the ones who make a brighter day.
  • The journey ahead is ahead.
  • What are you penning today? What adventures are you crafting by your doing? Get up, get moving… keep writing.

Look, it’s no ChatGPT, but it blew my mind! I’m only using a 4 layer NN with 7 million parameters. In comparison, ChatGPT uses 96 layers and 175 billion parameters! Before the weekend ended, I set up nanoGPT to build a more elaborate model on my data set. It’s still running, but already I can see it has learned a lot more of my style but seems to lack some focus on topics. It’s easily distracted and interrupts its own train of thoughts with new ideas. Squirrel! Nothing like me.

So my JasonGPT won’t be writing my Monday updates anytime soon, but who knows, maybe it will help me come up with some new ideas. I just hope it stays benevolent and kind. I would hate for it to suddenly become self-aware and start…

Connection to imac.local closed.
Posted on

Generative AI

Lightning across a digital eye of a typhoon

Typhoon warning! My nephew is a Lt. Commander in the US Navy currently stationed in Guam. He teaches and manages trauma and emergency care at the hospital. Last night, he was preparing his family for the typhoon that would be sweeping across the small Pacific island in just a few hours. They closed the storm shutters, stored their Jeep in the basement and ensure their backup power and pumps were working. My nephew drew the short straw at the hospital and will be managing the ER while the storm rolls through. I worried about the hospital being built for these type of events and he assured me that it was, but of course, he was quick to add that the generators were built by the lowest bidder.

There is another typhoon coming. Gazing out over the technology horizon we can see a storm forming. But this one seems to be more than heavy winds and rain. I’m talking about the recent astonishing developments in generative artificial intelligence (GAI). I’m increasingly convinced that we are sitting on the edge of another major tectonic shift that will radically reshape the landscape of our world. Anyone who has spent time exploring OpenAI’s ChatGPT or Dall-E, Google’s Bard, Microsoft’s Bing or Co-Pilot, Midjourney, or any of the hundreds of other generative AI tools out there, will immediately recognize the disruptive power that is beginning to emerge. It’s mind blowing. GAI’s capacity to review and create code, write narratives, empathetically listen and respond, generate poetry, transform art, teach and even persuade, seems to double every 48 hours. It even seems that our creation has modeled the creator so well that it even has the uncanny ability to hallucinate and confidently tell us lies. How very human.

I have never seen a technology grow this fast. I recall the internet in the late 1980’s and thinking it had the amazing potential as a communication platform. Little did I realize that it would also disrupt commerce, entertainment, finance, healthcare, manufacturing, education and logistics. It would create platforms for new businesses like the gig economy and provide whole new levels of automation and telemetry through IoT. But all of that took decades. Generative technology is announcing breakthrough improvements every week, sometimes every 48 hours. To be fair these large language models (LLMs) are all using decades old research in neural network (NN) technology. However, when you combine those NN with enhancements (e.g. newer transformers, diffusion algorithms), hardware (e.g. GPUs) and rich data sets (e.g. the internet) they unleash new capabilities we don’t even fully understand. The latest generations of the LLMs even appear to be doing some basic level reasoning, similar to how our own organic NNs help us solve problems.

Businesses are already starting to explore the use of this technology to increase productivity, improve quality and efficiency. Wendy’s recently announced that they are partnering with Google to use GAI to start taking food orders at their drive-throughs.1 Gannett, publisher of USA Today and other local papers, is using GAI to simplify routine tasks like cropping images and personalizing content.2 Pharmaceutical companies like Amgen are using GAI to design proteins for medicines.3 Autodesk is using GAI to design physical objects, optimizing design for reduced waste and material efficiency.4 Gartner identifies it as one of the most disruptive and rapidly evolving technologies they have ever seen.5 Goldman Sacks is predicting that GAI will drive a 7% increase in global GDP, translating to about $7 trillion!6

It’s time to prepare for the typhoon. I’m excited about the future! As a technologist, I know disruptions will come, challenging our thinking and changing how we work, live and play. I know it can also be terrifying. It can prompt fear, uncertainty and doubt. But now is the time to prepare! Don’t wait to be changed, be the change. Start exploring and learning. I have a feeling that this new technology will be a 10x amplifier for us. Let’s learn how we can use it, work with it and shape it to be the next technological propellent to fuel our journey to a greater tomorrow!

This blog text was 100% human generated but the image was created with OpenAI Dall-E2.

  1. Wendy’s testing AI chatbot that takes drive-thru orders. (2023, May 10). CBS News. https://www.cbsnews.com/news/wendys-testing-ai-chatbot-drive-thru-orders/
  2. Publishers Tout Generative AI Opportunities to Save and Make Money Amid Rough Media Market. (2023, March 26). Digiday. https://digiday.com/media/publishers-tout-generative-ai-opportunities-to-save-and-make-money-amid-rough-media-market/
  3. Mock, M. (2022, June 7). Generative biology: Designing biologic medicines with greater speed and success. Amgen. https://www.amgen.com/stories/2022/06/generative-biology–designing-biologics-with-greater-speed-and-success
  4. Autodesk. (2022, May 17). What is generative design? Autodesk Redshift. https://redshift.autodesk.com/articles/what-is-generative-design
  5. Gartner, Inc. (2022, December 8). 5 impactful technologies from the Gartner emerging technologies and trends impact radar for 2022. https://www.gartner.com/en/articles/5-impactful-technologies-from-the-gartner-emerging-technologies-and-trends-impact-radar-for-2022
  6. Goldman Sachs (2023, May 12). Generative AI could raise global GDP by 7%. https://www.goldmansachs.com/intelligence/pages/generative-ai-could-raise-global-gdp-by-7-percent.html