If you're reading this on a desktop, we suggest you open the PDF-version, but if on mobile phone - continue reading below.
Up until just a couple of years ago, it was more or less unfathomable for most people to see themselves having an almost human-like conversation with AI. Most of us thought that this would eventually pan out in far the off future, at least a decade away. Now it’s already here and quickly becoming a commodity. Simply by opening up ChatGPT, we now have some serious brainpower right at our fingertips. And doesn't really raise any eyebrows anymore. At least for people engaged in the digital realm.
On an historical time scale, what has happened with AI over the past decade is nothing short of spectacular. But we’re still only scratching the surface of what is possible. Our view is that, when we’re sitting around the AI campfire in a couple of years from now, we’ll be able to say the exact same things once more; that we couldn’t really grasp what was about to arrive.
As we’re climbing up the levels of AI capabilities, we’ll not only converse with an AI that uses Large Language Models (LLMs) and predict the most probable next word in line, based on a neural network stored in a black box backend system. We’ll be able to clone all of the information we’re in touch with, store it in a very defined and secure backend, and have it interact with the outside world in a plethora of ways.
Our prediction is that, over the course of the coming two years, we’ll all be able to create our very own “Jarvis”, if you remember the extremely capable computer system interface of Tony Stark from the movie Iron Man. Actually, the name Jarvis stems from the butler of Tony Stark’s father, Howard Stark, and the Stark household. Tony Stark created Jarvis and the attached acronym of “Just A Very Intelligent System”. So, it’s both a name and an acronym. Merging the old with the new...
Listen on Spotify:
So, what do we mean that we’re all going to create our own Jarvis?
We mean that we’ll be able to originate an AI creation that connects to all outstanding relevant information and have it become our butler (a super intelligent one) in whatever it is we want it to do. It will give us advice in how to plan and take action, no matter the situation or profession, and make us extremely more productive than we are today. It will know, for example, that there are two very important tasks you need to take care of today, or it’ll be challenging to meet a certain goal you’ve set out for next Wednesday.
It’ll be able to connect everything into the best possible action.
Sources for images: Marvel.
JARVIS IN INVESTMENTS
In terms of investing, having your own Jarvis that constantly brainstorms with you, offer new insights about events in real-time, and finding investment opportunities hidden deep in your dataset – will become a game changer for the industry. Investors will connect all their data they have, and I mean all their data, with everything from company-, industry-, and technology reports, to meeting notes, podcasts, and even research papers from universities.
Everything and anything will be gathered in a place where an individual investor will have total control and privacy, and with their datasets they’ll create Knowledge graphs with billions of nodes and connections that hooks up with an AI model and produce any outcome they want. And importantly, the outcome they want will be referred to as an unquestionable truth about a topic, not only convincing but incorrect statements made by LLMs. Hallucination will be eliminated and enable responses that we can fully trust and take action upon.
COMBINING KNOWLEDGE GRAPHS WITH LLMs
In general, and without getting too technical, Knowledge graphs provide structured and explicit knowledge grounded in factual data sources. They capture actual relationships, have greater truthful reasoning capability in many use cases, and are substantially more modular compared to LLMs. However, they are quite costly and complex to create. LLMs on the other hand, are trained with massive datasets and can reason and replicate human writing in way that’s more less magical. They can be immensely creative and offer things a Knowledge graph could never do.
But when they do this, it also creates reasoning that simply is not true. It drifts far from the harbor of truth. Hallucination, in other words.
Knowledge graphs structures information and creates clusters of nodes. Source: Neo4j.
When you combine the immense isolated potentials of Knowledge graphs and LLMs, and integrate them into a unified system that helps each other, you get the generative semantic power of LLMs with the accuracy and modularity of Knowledge graphs. Together, they’ll form the foundation of creating your very own Jarvis that is almost indistinguishable to humans in terms of conversation, and almost as truthful as mathematical equations. It will derive insights from your own dataset that you can trust, and you’ll know that your dataset will not end up in the hands of others.
So, just to repeat this upcoming and revolutionizing personal Jarvis creation that is coming – by fusing LLMs into Knowledge graph nodes and workflows, we’ll have supercharged LLM capabilities that are augmented with actual truthful data and able to reason (almost) like a human being. They are complementary to each other and improve the sum of the parts significantly.
For example, the creation of a Knowledge graph from an industry report is complex and can take time, because the information from the report needs to be categorized in a very thoughtful way so that all the nodes (containing information) are connected to each other in the most truthful of ways. One of many beauties with LLMs, is that they can process an industry report, meaning carving out all the individual pieces of information and create these nodes and connections, in a very effective way. And by effect, the LLM can use this web of nodes and give you factual insights with powerful semantics. Win-win.
THE THREE-BODY OPPORTUNITY
Let’s now move away from the technical and get more practical. Less technology, more UX. As an investor, sharing data is a delicate matter, while at the same time, using AI is becoming an essential matter. These two are in contrast with each other. If you use ChatGPT, or other types of LLMs, the input data could wind up in the hands of others. That’s unacceptable. While not using is also unacceptable, because we all want to save time and gain insights.
What we believe will pan out soon enough, is the ability for investors to open their own secure database (or data room) and have AI talk to this dataset. Now, this is not a rocket science projection, however, it is more difficult than you’d imagine, because you need to make the dataset into Knowledge graphs so that AI can retrieve and abstract relevant and actionable insights. This three-body problem is what we’re building at Sircular, and we call this upcoming product "The Second Brain".
Introducing Sircular's upcoming product; the Second Brain. It combines knowledge graphs with AI.
The combination of data, Knowledge graphs, and AI, will enable anyone to create and employ their very own Jarvis that constantly analyses reports, notes, podcasts, investment theses, and more – and become sort of like a super intelligent Warren Buffet in combination with Sam Altman. Powerful investment experience with the technical prowess of AI.
Investors will become as capable as Tony Stark (but without the saving the world-part...). They will wake up in the morning, go to work, have a coffee amongst colleagues, and then discuss the latest development in world with Jarvis, and create insights of what the potential impacts might be on existing or future investments.
And furthermore, if as this isn’t a mind boggling scenario enough, Jarvis will employ AI agents that constantly discover new investment opportunities and refine your investment theses, as well as integrating with other systems (like others' Jarvis) to retrieve the information you find most relevant into its Knowledge graph, allowing you to focus on deep thinking and research.
Jarvis will integrate the information you find most relevant, into his Knowledge graph and let you focus on the deep thinking and research.
So, in conclusion.
Just A Very Intelligent System, Jarvis – you’ll want to have one soon enough.
Until next time,
Christopher Lyrhem
Chief Future Officer at Sircular
Welcome to the Future is a research letter written by Christopher Lyrhem, Chief Future Officer at Sircular.
It gives investors, and other curious explorers, distillations of transformative future scenarios for our global society. Everything from autonomous vehicles, general-purpose robots, renewable materials, AI, circularity, and more, will be discussed. Sircular's mission is to empower investors with AI.
If you wish to test our Copilot (and very soon the Second Brain), visit our website: www.sircular.io
Opmerkingen