Welcome to the Exponential Age
Jul 26, 2023
Keynote speaker, Aaron McDonald, CEO of Futureverse spoke about a number of initiatives and AI-driven functionality that is already being used in the commercial sector to generate revenue, reduce costs and understand risk. Perhaps the most exciting of these is the marriage of the technology that drives the metaverse and the capability of AI to rapidly process and learn from vast amounts of data.
As Aaron explained, the metaverse is a virtual world where people can live, work, shop, learn and interact - all accessed through the internet. It has been around for many years and is probably best known for its application in online gaming (currently valued at US$249.55 billion). However it is much, much more than that with many believing it will deliver the next chapter of the internet - a single, shared, immersive, 3D world that sits alongside our physical world. Virtual Reality headsets and Augmented Reality glasses are just the beginning and are already being used in medical training. Digital Twins are being used to create copies of businesses in the virtual world where they can safely test strategies and tactics. Intelligent Avatars can be used as 'next level' chatbots to improve customer service and Intelligent Objects enable businesses to create potential new products in a virtual environment and get real time feedback from the market. For businesses in New Zealand, this is a space that we must monitor as it will impact how we do business and what consumers want and expect.
Britney Muller provided a timely reminder of the power of data to drive business decisions and shape our world. Indeed it is existing data (images, text, meta data, code) that is being fed into Large Language Machines (LLMs) like ChatGPT and Bard as a training base. In stark contrast to search engines (like Google), LLMs are fed data (from the internet, social media, books, articles etc) and use this to generate responses. They become increasingly more accurate in their responses when the user provides feedback which is then incorporated into the machine's 'brain'. Of course, it is not surprising to learn that LLMs are inherently biased because the data they are fed was initally created by human beings - who are often either intentionally or innocently biased. Further biases can occur during the manual process of 'cleaning' the data fed into the machines, in order to remove profane, violent and inhumane material.
Britney further explained that LLMs produce responses based on probability i.e. if you type in the sentence, New Zealand is best known for its...., the LLM searches for all the words that appear in its 'brain' and return the word with the highest probability. However, just to keep it human, there is a deliberate element of randomisation built into the model!
With this interactive and iterative response behaviour in mind, along with the inherent biases in the data it is important to remember 2 key factors:
(1) all responses should be fact checked and
(2) never enter personal data - you never know where it might pop up again.
Of course, LLMs are just one form of AI, with OpenAI's ChatGT being only one example. However with its natural language capabilty, it's not surprising that ChatGPT hit the headlines, attracting 1 million users within the first 5 days of launching and as at January of this year, have 100 million users.
So what does this all mean? It is clear that we have reached a tipping point in technology and we can expect some fairly significant disruption from business-as-normal as we go through the next few years. COVID19 has taught us that vigilance, agility and pragmatism are our most valuable mindsets and that should stand us in good sted as we enter the Exponential Age.
