Introduction

A number of discoveries in deep learning technology in the mid-2010s led to new architectures and highly parallelizable model training using internet-scale data. In the 2020s, newer and larger versions of diffusion and transformer models further increased adoption. This was made possible by the ease of improving the models through fine-tuning, natural language prompting, and increased open-source activity (LLaMa and Stable Diffusion, among others).

These trends collectively led to a broader adoption of generative AI-powered applications, reaching 100s of millions of users by the end of 2023. However, we believe that we are only beginning to see the economic impact of these technologies. cyber•Funds aims to support founders, researchers, and product teams who are working on two main objectives: (a) building new AI-powered capabilities and (b) bringing AI products to every area of the economy.

The mission of cyber•Fund is to enable the cybernetic economy. AI plays a crucial role in the programmable economy and has the ability to significantly reduce the costs of cognitive tasks, ranging from personal assistance to coding, art, education, healthcare, and even scientific research.

What is the mission of machine learning?

Human progress has an upper bound in the form of complexity and communication bandwidth. Our wetware RAM can only handle a limited amount of complexity, and human communication has its limits in terms of information flow. However, advancements in language models (LLMs) and other models solve this problem by allowing us to outsource cognitive tasks to software. This enables humanity to deal with the increasing level of complexity and information flow.

Throughout history, technological progress has always led to automation, whether it's the printing press, railroads, steam engines, or the internet. However, the progress itself depends on cognition and intelligence. By automating intelligence, we create a positive feedback loop in technology innovation. By enabling a cybernetic economy and accelerating the rate of innovation, we open ourselves to the ability to solve every other problem or need that our civilization has opening up the pathway to post-scarcity society.

Goals

The purpose of this document is to present an investment thesis for cyber•Fund for 2024 with the focus on Artificial Intelligence technology and aligning with the overall vision of the cybernetic economy.

In this document we examine promising yet undercapitalized themes within AI vertical. We start with mapping the market, identifying core investmenet themes. We proceed with defining the strategic priorities to focus on in 2024. Finally, we derive a list of action items and open questions.

Growth opportunities

AI is generally useful for any task where automation or enhancement of cognitive abilities is required. Thus we believe that AI will be the leading driving force for the growth of technology sector and economy in general.

On a high level, growth opportunities in AI arise in two dimensions:

  1. Horizontal growth: adaptation of existing technologies to larger market. We believe that even the current generation of LLMs is still only being used by a small percentage of potential addressable market. The value will mostly be created by operationalizing AI usage, innovating in UX, distribution and broadening areas of product applications.

Screenshot 2024-01-09 at 14.48.05.png

  1. Vertical growth: creating new capabilities, architectures and functionality. We are still ongoing rapid innovations in AI, and the scaling opportunities are not yet showing signs of diminishing returns. This means we will likely see more powerful, capable, efficient and overall more useful models and application being created in the coming months and years.

Areas of interest

In addition to purely AI application, we are looking the intersection of AI with

  1. Crypto facilitates decentralized computation for AI model training and inference.
  2. Web3 and blockchain offers a decentralized, transparent economic framework for AI agent operations.
  3. Self-sovereign identity and distributed data store systems guarantees data privacy and provenance in AI model training and application.

Strategic Priorities