?? AI is the real Web 3

Large language models like ChatGPT and GPT-4 are the future of the web.

Over the years since it was first conceptualised in 2004, Web 2.0 has transitioned from a social web to one ruled by major online aggregators like Facebook. For the past few years, the search for Web3 has been focused on building a data model on the blockchain that focused on individual self-sovereignty. Despite its promise, adoption has been slow, and developers in the Web3 space face regulation challenges due to some dubious claims and behaviours.

In contrast, AI’s rapid improvement has led to the emergence of large language models (LLMs) and new behavioural interactions across the internet. These models have immense potential to revolutionize how we access and use information online. For example, even though products like ChatGPT and GPT-4 are still in development, they are already being used more than blockchain-based Web3 products for practical day-to-day purposes. I have gotten more done using ChatGPT than on the Ethereum Virtual Machine.?

OpenAI’s recent announcement of “plugins” to connect chatbots with underlying services like Wolfram Alpha, Kayak, or Shopify makes LLMs even more powerful. These plugins provide a conversational interface to trusted information, mitigating weaknesses inherent in LLMs, such as hallucination or insertion of random information. Microsoft grounds LLM outputs to your data in its forthcoming Office Copilot to increase the robustness of its models. This technique is likely to be similar to approaches used on-prem/behind-the-firewall LLMs to make enterprise LLMs more reliable.

As LLMs and the systems in which they are implemented improve, these natural language interfaces will become increasingly capable—and help us get real things done.?

Interfaces driven by LLMs could define the future of the web. However, the question remains whether there will be a single dominant player like Google in Web 2.0 or a more fragmented landscape like social media. I believe we will see multiple services for numerous reasons. Organizations and countries may prefer to maintain control and privacy over their data by using their own localized or on-premise systems. Additionally, privacy-safe, on-device LLMs could grow in popularity to avoid data harvesting and privacy concerns associated with online services like ChatGPT and Facebook.

While it’s uncertain whether a single dominant player will emerge, it’s clear that the future involves LLMs connecting to trusted sources and services. These models will likely change interaction patterns and business models in a way that will define the successor paradigm to Web 2.0.?

(The full video is available here.)

Follow me on LinkedIn.