Skip to main content

2024: The year we scale generative AI

My inbox and social media feeds are currently filled with predictions for 2024. The next 12 months will be, in no particular order, the year for: multimodal AI, video generative AI, AI regulation, distributed training, AI agents, AI robotics, low-cost generative AI, edge generative AI, sustainable generative AI, and so on.

These predictions are all really useful for understanding the overall direction of travel for AI for the year ahead, and pre-empting some of the new features and products that will be developed this year.

However, these predictions also tend to focus on technological developments, and fail to capture some of the equally important macro level trends that are happening across industry and society more broadly.

For me, 2024 will be the year when we start to really test the scaled commercial value of generative AI. And whilst it’s not a slam dunk, I remain hugely optimistic and excited for the road ahead.

Moving beyond proof-of-concepts

Throughout 2023, we saw a huge amount of interest and activity around generative AI across all sectors, mostly focusing on generative AI strategy development, use case identification and proof-of-concept (PoC) testing.

These are all hugely important activities when a new technology matures into the wider market, but they are only the first steps in achieving the business outcomes that most organisations typically seek, namely growing revenue and reducing cost.

In 2024, we’ll see many more organisations attempt to bridge this gap, moving from isolated testing and boardroom strategy to deploying productionised generative AI models and products at scale into frontline sytems, processes and teams.

Scaling is always challenging

This sounds simple enough, but as many readers will know from previous experience, scaling is always more challenging than expected, and scaling generative AI is no exception.

There are a whole host of challenges that organisations must overcome to achieve sustained value from scaling generative AI. This not only includes identifying and prioritising the right use cases, but also effectively navigating buy vs. build decisions, managing upfront and on-going costs, balancing and mitigating risks, integrating new solutions into existing systems and processes, evaluating model performance for a diverse and complex range of tasks, ensuring sufficient wide-spread adoption, managing rapid and on-going developments in AI, and more.

There are also arguably more fundamental scaling challenges – and opportunities – around organisation-wide transformations. Generative AI doesn’t just enable organisations to do the same things quicker and cheaper, but provides opportunities to fundamentally change the ways in which organisations operate at scale both internally and within the wider market. Scaling is not just about using generative AI across an organisation, but also managing strategic and organsiational changes that are enabled by the technology.

2024: When the rubber hits the road

Will organisations succeed in progressing from strategy and experimentation to sustained, scaled value from generative AI? And will generative AI deliver on its promise and hype of 2023?

My prediction is partly yes, although I also expect deeper, more sustained value will be much harder won and accessed in the shorter term only by the most innovative and effective organisations.

More specifically, I suspect many organisations will succeed in integrating generative AI somewhere into their existing operations, mostly in simpler use cases where generative AI is clearly useful out-the-box and where the return on investment is obvious upfront. Examples include generating images and videos for social media marketing, proof-reading and copy-editing text, summarising documents to support decision making, accelerating technical development through coding copilots, and so on.

I see a lot of use cases tackling these quick wins, and they’re certainly a useful place to start. I also see an increasing range of off-the-shelf products that can be used by organisations to help solve these use cases, and provide organisations with the security, privacy, customisation, integration and support that is required in commercial applications.

However, I think it will be much more challenging for organisations to successfully achieve deeper, more sustained and transformational value from generative AI. This is much harder to do, not just technically, but also at a strategic and organisational level. It requires organsations to not only fundamentally rethink the way in which they do business, but also to have the ambition and buy-in to tangibly go after these opportunities, and the skills and experience to do so effectively.

This is what I see Deloitte itself working on in real-time, and is also where I expect the more innovative and ambitious organisations to focus in 2024.

Success in this latter area is certaintly not a foregone conclusion, at least in the nearer term: there remains huge uncertainty with generative AI, especially when exploring more complex applications where the value remains unproven, where there are no existing off-the-shelf solutions, and where the cost to unlock this value is unknown and potentially high.

However, this is also where I see the greatest value from generative AI, both in organisations and in industry more broadly. We’ve seen similar transformations unfold with previous waves of technology, including the advent of the internet, the smartphone, social media, and cloud computing. It’s not always smooth sailing, but these technologies have fundamentally changed the way in which we all operate.

2024 will be the year when we start to test if and when generative AI will do the same.