“If you thought the speed of change was fast this year, 2024 will continue that trend and the pace will likely increase.” David Elliot, AWS’ Head of Solution Architecture
If 2023 was the year of waking up to the opportunity posed by AI, 2024 is likely to bring significant changes for how enterprises approach data and AI. Some popular use cases we see now may become obsolete and the focus on improving efficiency may be replaced by more nuanced and bespoke ways of using AI.
The emergence of the EU AI Act is likely to dominate the agenda, but our in-house technology experts have given their predictions for the year ahead, based on their first-hand experience of implementing data and AI solutions within highly regulated enterprises. We’ll check in at the end of Q2 to see how accurate these predictions have been…
Companies need to adapt how their business enablement functions support AI assisted ways of working and technology. So how technology, governance, risk, compliance, legal, finance need to adapt to drive AI adoption in the enterprise. Lots of key dependencies haven’t been met to drive broad adoption of AI at scale. Those who are already doing this are set up for full scale transformation, those who haven’t are at risk of AI becoming a siloed technology instead of an organisation-wide capability.
We mentioned before how users, beneficiaries and stakeholders are not involved in the development cycle early enough, leading to a lack of understanding of the impact of AI. As such, organisations this year will begin introducing the right level of support for scaling these new systems, but may find this a slow process due to lack of expertise and platform level capabilities to actively support them beyond very small use.
Companies will also realise that just plugging in solutions such as OpenAI / ChatGPT won’t provide any real differentiated offering to their customers. If it’s easy, then everyone will do it. It’s far more cost effective and valuable if you break down bigger problems into a group of smaller problems and understand where specific AI services can solve those smaller problems individually.
The best solutions are simple ones and companies should avoid applying AI / machine learning where much simpler solutions will resolve the problems. Many problems we see with our clients could be solved by simply changing an existing set of processes and rearranging how teams organise themselves.
More organisations will use Generative AI to govern how they use Generative AI. We’ve seen demands from organisations with highly in-demand data and enablement teams already, who are taking their internal controls and policies for AI and codifying them into a bot. Then these teams can ask the bot how they need to govern and control their use of AI, giving you the starting point for the controls around your use case.
For enterprises who have more use cases they have the bandwidth to look through, this could open up a path to innovation. Organisations we’ve spoken to have 1000s of AI use cases in the backlog and people wanting to use Generative AI, but no capacity to build the relevant systems and technology.
As this demand outstrips regulatory controls, we could be in a position where people are frustrated with the bottleneck of centralised IT teams. This leads to them going their own way, adopting technology outside of controls and policies and the threat of shadow IT.
While 2023 was the year businesses woke up to the importance of data quality, I think 2024 will be the year the enterprise connects the dots between a data-driven approach and business value.
By anchoring a data mesh approach to concrete business outcomes that go beyond minor tweaks to systems and processes and result in real change, organisations can begin to derive real value from their data. This will create the usual camps of trend setters and laggards, which we’re already seeing across different industries with some catching on a little quicker.
As the adoption of AI continues at its current breakneck pace, the importance of data quality will rise in parallel. But those with mature data strategies and a well defined data mesh approach will be able to capitalise on opportunities within their own data and leverage AI models to extract even further value. The two camps will grow wider, and we’ll see tooling offered to fix up data estates when really it’s the strategy and the approach towards data that needs reimagining.
The importance of partnerships, and what cloud providers are able to offer enterprises, will only continue at its current rate of expansion. Generative AI has rightly dominated this space, but offerings are likely to become more nuanced and give organisations the ability to stack technologies and apply models to their own data.
Mesh-AI's strategic focus on becoming the AWS no. 1 partner for data & AI is a strong indicator of its commitment to deepening its partnership with AWS. With its existing collaborations and demonstrated expertise, Mesh-AI is poised to achieve Premier partner status, having now been awarded as Sustainability Partner of the Year 2023. Gaining AWS Premier partner status will solidify Mesh-AI's reputation as a leading player in AI solutions. This achievement will likely open doors to larger and more complex projects, further boosting the company's growth.
My view is that organisations and architecture teams will stop debating whether the solution to their data problems is a data mesh, data fabric, data mart, data lakehouse. Instead organisations will take a more pragmatic approach that looks to combine the best of all of these concepts for one main goal: how do we make high quality data available across the organisation to deliver value to the business.
Part of this is going to be fueled by the fact that exec and senior leaders will be under immense pressure to deliver value with AI. And with off the shelf tooling from cloud providers reducing the barriers to AI adoption, architecture and data teams won’t have the luxury for dogmatic debates.
The other reason why I see this being the case is vendors will increasingly start providing modern data stacks that blur the boundaries between these concepts, and therefore organisations will have no option but to adopt them, especially if they want to capitalise on the out of the box AI capabilities that come with it. Technologies like Microsoft Fabric are increasingly going to push organisations to focus on the matter at hand – namely making good quality data available – and spend less time on thinking about what's the perfect archetype of a data platform/ecosystem. I also believe that each of these concepts bring something unique to the table, and a combination of these would serve organisations well.
I’m predicting that the value will run out for isolated use of Generative AI. Prompts have become more complex but there appears to be an end to the runway here. However, combining Generative AI with other types of machine learning will make predictions based on input data more accurate and allow businesses to do far more in this area.
This stacking of technology will enhance the output of Generative AI but also reinforce the importance of the underlying data. Those who can curate the data more effectively and point AI to specialised tasks will win.
Finally, given the number of enterprises with ongoing or established AI operations, we are likely to see them fine tuning existing models and activities to make them more effective and delivering value. Given this, it’s likely we’ll see frameworks for this sort of fine tuning democratised by cloud providers.
.png)
.png)
