“You can dream anything” — a lie “You can dream anything” — a lie It is getting scarier by the day to imagine the world that is predominantly dependent on AI. Everyday there is a newer and dystopian version of the future that is being portrayed by the media. How true is it ? I am completely convinced that AI is better at writing code than humans now. I was reflecting on last week’s work and I didn't reach out to the code in IDE once in the last two weeks. I was working completely out of the terminal switching between the OpenCode and terminal prompts and I was able to get the Agent to get me exactly what I wanted without going too much into the specifics of the implementation. This is a huge leap from what was possible just months ago. What about other jobs now ? Is the LLMs creative enough to replace human creativity? Are we still the steward of the creative direction of the products we are working on ? LLMs are not creative LLMs are not creative AI is coming for our jobs vs AI can never replace the human creativity. There are people on both camps. Some are saying AI is a tool and you can only use it to get results. While others are scared that AI is capable of doing much more that there will be fewer and fewer jobs as it gets better and better. The camp that believes AI is not creative enough is basing it on the following facts about LLMs. LLMs do not Reason , they Rationalize. LLMs dont Ponder, they Process. LLMs cannot create Endless information. LLMs do not Reason , they Rationalize. LLMs dont Ponder, they Process. LLMs cannot create Endless information. Simply put AI Agents are basically regurgitating content. AI Agents/LLMs are a really complex autocomplete system that just predicts the next word that needs to be generated and It gets better at predicting by repurposing all the patterns it has encountering so far. But I would argue that humans and LLMs are the very close to how they operate and how creative they get. Dont get offended, Please hear me out. One of the biggest lies we tell ourselves is that we can dream of anything and achieve it. Humans can dream. But can humans dream of new information they don’t know already ? The truth is, we can’t. Humans can dream of new places and new structures. However, it is always based on a permutation of things and experiences already lived through by the person in real life. You would never hear a word uttered in a dream if you haven’t already heard that in your real life. We are creative in our dreams, but only to the extent of rehashing experiences familiar to us. All the music in the world is based on only 7 notes. All stories are based on 7 plots. All the music in the world is based on only 7 notes. All stories are based on 7 plots. Take music for example. We feel musicians and artists are infinitely creative. But all music is a play on different combinations of arrangements on these 7 notes, only the rules of how to combine them change. The rules on Beats (Thala) , Tempo ( Gati ) , allowed notes for the song ( Raaga) and the instruments determine the genre of the song you listen to . Humans have created an endless music library already that will take many more generations to explore completely. And we will create even more as we go forward. All the stories and movies that have been created ever can be reduced down to 7 plots. reduced down to 7 plots If possibilities are infinite with 7 notes and 7 plots, imagine the possibilities of an AI Agent with an LLM that has been trained on all the text in the world till date. A movie director makes movies on experiences he has lived or heard. A visual artist is usually addressing a pain or concern that they have encountered, or putting together different mediums they have experimented with. A rapper is usually putting rhyme to words that describe his life or community. Humans aren’t the only creative creatures. But Humans are the most creative creatures we have encountered so far. But is all human creativity new and infinite ? Human are an Autocomplete machines of their lived experiences. And Creativity happens when we connect the dots in our knowledge , life and experiences. Human are an Autocomplete machines of their lived experiences. And Creativity happens when we connect the dots in our knowledge , life and experiences. connect the dots Humans relate to each other through shared experiences. Internet meme culture has taught us 2 things, you enjoy the meme when you relate to it. But most importantly , you can only make a meme of a popular concept for virality for that same reason of being able to relate to it. A meme is nothing but a person applying a new situation with a familiar situation from a movie, media or real life incident. People are quoting the same dialogues from your favorite LOTR, Harry Potter or Marvel characters. You can classify 99% of the memes in this bracket. Some people are extremely creative with a completely new concept, but they are rare. LLMs cant match humans like Newton and Einstein LLMs cant match humans like Newton and Einstein Newton did discover gravity. He had been pissing towards the ground all his life, but It took an apple hitting his head for him to realize that there is gravity . This was a lived experience. Steve Jobs learnt calligraphy and Typography. So he really cared about Graphics, which led to the choices he made to make graphics important on the systems he worked. This was a lived experience. Einstein worked in the patent office where he was experiencing a lot of patent ideas that came from other peoples lived experiences , giving him the freedom to ponder and put together much more complex concepts from what he learnt. This was a lived experience. I am not belittling the work of these great people. There is a reason we gravitate towards these names, and that is they are outliers. They are people who pushed boundaries of human capabilities and imaginations. They belonged to the 1% of the crowd. What about the 99%? All the rest of us can only think of solutions from our past experiences. Atleast that is the solution you can think from the top of your mind in this distracted world unless you set time to ponder. Isn’t LLM better at arriving at the top of the mind solution ? they are outliers Basic building block of LLM ( the brain) and RAG (the context or memory) , in simple terms is the ability to build association between words and concepts. In other words , Connecting the dots. As LLMs are fed more and more information about different domains. They can quickly draw connections and add up the information . AI Agents well versed in Legalese can talk to an AI Agent with Economics knowledge to draft new recommendations on social order. They don’t need to reason or ponder, they only need to process and rationalize. and they can be creative without pondering and reasoning. Connecting the dots. They don’t need to reason or ponder, they only need to process and rationalize. and they can be creative without pondering and reasoning. Good artists copy. Great artists steal Good artists copy. Great artists steal Creatives hit a creative block often. It is hard to think of new ideas everyday. That is why books like “How to steal like an artist” exist and teach us know to convert inspiration into creative work by applying your personal flavour and imagination. Most work is an extrapolation of existing work. Reels and TikTok are great examples of the same rehashed content being applied by a million times. Dancing the same steps done by million others is hardly creative. “How to steal like an artist” LLMs dont have creative block when it comes to inspired work. They have consumed more inspired content that you can even imagine as possible. In AI generated content, AI slop is just generated by people who are not creative enough. True creatives are excelling and getting empowered by the use of AI. But a day is not far away, when an Agent trained on movie scripts can instruct an AI Agent capable of 3D animation to pull off a full length feature film. True creatives are excelling and getting empowered by the use of AI. True creatives are excelling and getting empowered by the use of AI. AI Agents are better at knowledge work than 99% of the people. AI Agents are better at knowledge work than 99% of the people. AI Agents don't have free reign yet. They are just tools . But they are tools that are better at executing instructions than humans. They won't replace humans ( yet ) , but they augment your work and make it faster to achieve things. They without doubt reduce the need for humans to execute. We need to brace for an eventuality where every workplace is a ( much ) smaller version of what we have today. 10 marketers become 2. 10 engineers become 1. We need to find purpose for the vacuum created for 80-90 % who are left behind. AI Agents work 24 x 7 and are relentless in trying as many possibilities needed to arrive at a solution. We will experience fatigue from too much content being generated with the help of AI and move to handing over most of our decisions to Agents because it wont be humanly possible to keep up with the updates. Content will nosedive in value soon. AI Agents are as good as humans at code and at knowledge work. I have no doubts and I insist that you don't underestimate the possibilities either. We will find a synergy after the chaos , but we need to first wade through the storm for the rest of this year. Best thing we can do is use AI as much as possible and get shit done and wait for this to play out. AI Agents are as good as humans at code and at knowledge work. I have no doubts and I insist that you don't underestimate the possibilities either. AI Agents are as good as humans at code and at knowledge work. I have no doubts and I insist that you don't underestimate the possibilities either. Systems that existed for centuries are about to change. The ratio and causality between work and outcomes is about to change. Now I am off to wonder what that new world looks like. Maybe I can think of something that I don't know already about all this. Maybe not ! Systems that existed for centuries are about to change. The ratio and causality between work and outcomes is about to change. Systems that existed for centuries are about to change. The ratio and causality between work and outcomes is about to change. If you felt this article was creative enough, do give me a thumbs up. AI Agents don't need validation, but I do. :)
