The rise and rise of the AI co-pilot, as ‘Y2Q’ looms

Simon Bush

Over a week in late November, the Australian Information Industry Association led an trade delegation to the United States made up of digital businesses and senior government representatives.

The delegation visited 14 world-leading technology companies in New York and Seattle comprising federal and state tech leaders, Australian digital startups, medium businesses and multinationals.

There was a dominant theme of never-before-seen scale of investment from every company we met – that being generative AI and the rise of the co-pilot.

The discussion on generative AI fronted almost every presentation with responsible AI a core component. Not only are software companies investing many millions of dollars (or tens of billions by the cloud platform providers) each in their own large language models (LLMs), they are focused on doing so in a transparent and open way.

We heard from the hyperscaler cloud infrastructure providers about their investments, which are monumental, with the scale the likes of which have never seen before in our industry.

Image created by author using Adobe Firefly generative AI

For example, one global tech company is investing US$10b each quarter around its need for new data centres and compute as they expect their cloud platform and related AI co-pilots to drive this demand.

Another global tech company referenced the work it is doing to mitigate the massive energy costs of the new compute (GPUs) required to run LLMs and is investing US$15b over five years in renewables alone to power its data centres. The SaaS providers we heard from are each investing many millions to bring co-pilots and automation across their software.

A key message from tech companies is that large amounts of data are needed for LLMs to be insightful and useful (with corresponding large amounts of compute) so a major focus is getting large data sets from users (customers) and, interestingly, some companies indicated quality of data is a secondary but still important consideration.

The delegation also heard that interoperability of vendor AI and ML models is considered best practice, and that cyber security is critical in the design and roll out of the new technologies, with AI assisting with surface threat management and penetration detection at a scale that humans alone cannot achieve.

A future trend will be industry vertical LLMs being created due to the prohibitive cost for a single company creating its own LLM (and the need for larger data sets to make the LLMs more insightful and powerful).

One leading company indicated that a future area of innovation in AI is trying to understand behaviours and prompts. Numerous other companies are thinking laterally and investing in new skills such as hiring anthropologists and psychologists to assist in (ethical) product development.

We also heard that by 2025, 85 per cent of all software applications will be built assisted by AI. Accordingly, we may be close to the top of the AI hype-curve, but the longer-term impact is underhyped.

The secondary theme – also related to AI – is that of automation. Tech companies are heavily investing in capabilities that allow organisations to leverage the powerful automation tools that will unlock talent and raise productivity by streamlining business processes and bringing intelligence into the business software platforms.

We heard from one company that through automation had reduced developer monthly logins (and thus cyber security vulnerabilities) from 30,000 to just 90.

The focus for some companies was on quantum and the current and future threat that it poses to unlocking current asymmetric cryptographic measures.

An important learning was that encrypted data sets that have been exfiltrated today will be cracked by quantum computers in the future (hack now, crack later).

The good news, however, is that industry has risen to this challenge and that quantum cryptography can be leveraged now so that data sets will remain secure even after the advent of widespread quantum which is believed to be from around 2030 onwards.

Industry standards on quantum cryptography have also been developed and approved (only in the past few weeks).

The industry has coined the term Y2Q (like Y2K) with the year it is expected to hit us and be a threat by those (likely nation states) who wish to use quantum for ill is 2033.

The US government in 2021 invested US$1b and two years later in 2023 it has invested $24b, China is investing $10b. Australia’s quantum investments – along with AI – is too small and too slow.

The US Government has also required all its agencies to report on post-quantum cryptography vulnerabilities that exist today, something that the Australian government may wish to consider.

A final comment on working from home. We are starting to see a reversal from US tech companies from 100 per cent working from home (WFH) to a compulsory hybrid model, with a minimum number of days in the office required.

As one executive concluded on WFH: it increased productivity via increased hours worked, but it decreased creativity and innovation. We are seeing this trend reversal play out in Australia across a number of industries right now.

The US tech companies have a culture of innovation and investing in new capabilities and inventing new models leveraging small cross functional teams of around 10 people (‘two pizza teams’) working together on projects over 18 months.

We can learn much from the culture and approach to innovation from these successful and innovative tech companies, where some have over 75,000 software engineers alone and yet continue to innovate despite this scale.

It was a fascinating time to meet with companies who have been part of the innovation and investments driving generative AI that are here with us now and will change our world.

We also received insights into how far tech companies have come in developing and integrating LLMs into their products and brought to market in a short period of time, and we are only at the start – version 1.0 – of this fascinating generative AI journey.

The delegation provided numerous benefits including government to government and government to industry engagement and the importance for Australian-based companies and officials to see the latest global tech developments in action.

We are at the beginning of the AI revolution in many ways with challenges of Quantum now emerging. Australia must learn and innovate to ensure we have our own place in this story.

Simon Bush is the Chief Executive Office of the Australian Information Industry Association (AIIA)

Do you know more? Contact James Riley via Email.

Leave a Comment

Related stories