24.03 Liontrust Global Innovation Report - The Rise of AI 04.24 - Flipbook - Page 34
AI – tried and tested by the Global Innovation team
Not only is AI improving productivity by enabling people to accomplish more with less,
but it is also helping open up new markets and stimulating growth.
Key to this growth is the concept of democratisation: by utilising
AI-powered technologies, people are being empowered to
engage in activities that were previously out of reach, ranging
from coding and app development to writing and creative
design. To test this idea out, members of the Global Innovation
team looked to experiment with various AI applications in a handson way – such as Microsoft’s Copilot, Adobe’s Firefly and Open
AI’s ChatGPT.
What Copilot means for Microsoft and us
“We believe Copilot will fundamentally transform our relationship
with technology and usher in the new era of personal computing.”
— Satya Nadella.
1 November 2023 was a momentous day for technology, for
consumers of technology and for investors in technology. It was the
day that Microsoft announced the general availability of Copilot
for the enterprise, its generative AI tool designed to enhance
productivity across computing tasks. Why was this so significant?
Two key reasons. Firstly, because Microsoft has so successfully
become the operating system of computing for the vast majority of
humanity, Copilot has the potential to meaningfully impact us all
in our day-to-day lives. Most of us use Microsoft in some capacity,
whether using Microsoft word documents, building spreadsheets in
Excel, using GitHub to develop code, or Bing to shop. Copilot is
being infused into all of these products – Microsoft is democratising
access to this new generation of technology. We are all going to
have access to our own generative AI assistants.
Secondly, the productivity gains being delivered by Copilot are
eyewatering. Data just released from customers of Copilot’s early
access program (who used Copilot for eight months leading up to
November) demonstrate that generative AI-driven productivity gains
are both real and, even more notable, realisable from day one of
use. 70% of Copilot users said they were more productive (by a
factor of 55% for developers), 68% saw an improvement in the
actual quality of their work, and 77% said once they used Copilot,
they didn’t want to give it up.
34 - The rise of AI: Technology and Innovation Report
Naturally, when faced with a customer value
proposition as striking as this, the Global Innovation
team wanted to test out Copilot ourselves. We
decided to build our very own website using
Copilot in Microsoft Power Pages,
which lets users rapidly give natural
language prompts to generate
relevant content, code, Cascading
Styling Sheets (CSS), and images into
a webpage, as well as embed their own
chatbots in their websites. However, we fell at the
first hurdle. Copilot in Power Pages has not quite made its way onto
our computers yet. Undeterred, we set upon building our website
using Wix’s ADI (Artificial Design Intelligence) tool, the company’s own
version of Copilot, which we thought would provide a similar enough
experience.
Concluding thoughts: ’more for less’
The ascent of low/no code development (such as developing this
website) is hard to overestimate in terms of its impact on developer
productivity, and this is being driven by generative AI. The average
developer spends approximately 50% of their time writing code,
but a 55% productivity infusion from coding assistants such as Wix
ADI and Copilot can save up to 2.20 hours per day, freeing up
their time for design and enhancing both the output and quality
of their work. If the Global Innovation team can create a website
using simple natural language prompts in under 20 minutes, just
think of the collective productivity gains that can be realised as
copilots and AI assistants accelerate their global rollout.
Firefly – a small step for AI, a giant leap for Global Innovation
In the field of creative image design, we decided to experiment
with Firefly, an exciting new generative AI model released by one
of our current portfolio holdings, Adobe. This model can recognise
connections between text and images to generate image outputs,
and now powers an array of tools across Adobe’s creative software
suite. Leveraging Adobe’s vast scale and IP, the latest version –
Firefly 2 – was trained on over 500 million images and videos
