Today, the possibility of leveraging Exponential technology to create societal impact at scale seems more attainable than ever before - a sentiment I had previously expressed. What is particularly intriguing is that Generative AI appears to be receiving an additional, invisible impetus in the form of 'Convergence'.

Historically, the AI landscape was characterized by distinct specializations including Voice, Image, Text, Robotics, among others. However, advancements made in one area often failed to positively influence the others, resulting in isolated advancements.

The advent of Generative AI technologies powered by Transformers has ushered in a new era of convergence where previously separate categories can now be unified. This new multi-modal paradigm allows text to generate new text, images to generate new images, and text to generate images and so on, creating a positive compounding effect where advancements in one category can benefit another.

This convergence has produced a double-exponential impact, driven by both the Transformers and the combinatorial innovations from multiple AI categories.

The progress we have seen is undoubtedly positive and one that we should all celebrate and embrace. However, it is imperative that we remain mindful of the potential downsides.

We are currently witnessing a surge in unintended consequences such as biases, hallucinations, deep fakes and more, which have also begun to exhibit exponential characteristics. It is hence crucial to develop systems that amplify the gains and minimize the loss, thereby creating a net positive exponential. I refer to this net difference as the "Exponential Delta."

Exponential Delta = Exponential Gains minus Exponential Loss

‘Societal Benefit’ occurs when Exponential Delta is positive

‘Societal Harm’ occurs when Exponential Delta is negative

We acknowledge our responsibility to develop systems that generate a positive "Exponential Delta" for the benefit of society. To accomplish this objective, we have established an array of capabilities, including:

Trust - Ensuring that the knowledge and insights generated originate from sources we trust, rather than the internet.

Provenance - Maintaining traceability to the sources of knowledge.

Telemetric governance - Providing transparency and a better understanding of the interactions that happen on the platform.

Feedback Loops - Establishing a process of continuous learning and improvement.

Credentialed users - Granting access to qualified and trustworthy users.

Bias Mitigation - Incorporating diverse sources and strategies to prevent bias.

By fostering a culture of openness and promoting collective wisdom, we believe that the true benefit of these capabilities can be realized. This involves recognizing inequalities and embracing diversity to create a platform that is truly mutual.

Photo by julien Tromeur on Unsplash