“92% of CEOs know they need to employ AI & Automation” is a piece of research carried out by EY in 2023. The statistic alone is unsurprising and predictable. As with many things in life, it is what is not stated that is more important. Part of the research covers concerns shared by senior leadership teams.
When you look at the research and what is happening in the wider corporate world, a clear pattern emerges. Here are the top 5 things that are holding companies back from progressing.
1. Lack of Skilled Workforce
2. Complexity and Integration Issues
3. High Initial Costs
4. Data Privacy and Security
5. Regulatory and Compliance Concerns
As is our style, let’s break it down and look at the underlying challenges and how they can be addressed. Read on LinkedIn
Lack of skilled workforce
This is remarkably common but is completely avoidable. Many new technologies are rolled out to users with either no user training or the basics. Today’s productivity tooling is incredibly powerful but still we see user bases that do not know how to use functionality that has been around for 15-20 years.
The rate of change in technical advancements today well exceeds anything we have seen before. The simple answer to this is not to look at AI & Automation as just another tool. As a senior leadership group, it is necessary to include strong training and education for your workforce. AI & Automation will change the way your business operates, and your users work. Increasingly we are seeing businesses training vast swathes of the user base, sometimes in excess of 40%, on Power Platform to allow users to resolve and automate their own work.
For Generative AI, the power of the prompt is often overlooked. If prompt engineering is overlooked, you will end up with users utilising only the very basic components. Whilst meeting recaps, minutes and image creation are fine, you will restrict business to a very low rate of RIO. To cover the cost of a Copilot for M365 license for a month, it only really takes a two hour time saving. Great, but hardly ground breaking. If you invest the time for users to learn how to use the tooling properly, and how to share that experience, you will see the ROI % rise to the hundreds and the payback time drop dramatically.
Complexity and Integration Issues
This is a genuine concern. Senior leadership teams will be pressured, either from the technical teams or the business to deploy solutions. With pressure coming from so many directions and for so many solution areas, where does a business start? Is it sales, operations, back office functions, CRM, B2C solutions and what are the risks in doing so for each?
Complexity is often actively introduced when working out that ‘big win, high impact, silver bullet that will revolutionise a business’. Is that the right place to start looking? Potentially, but probably not. A clear strategy for this is imperative as it allows for methodical and considered approach. Approach the big solution? Break it into manageable phases, or look around the business for the real value wins?
Utilise frameworks to establish where to go hunting, what the likely returns are going to be, and what the risks are, is the way to start. Don’t get led by the overenthusiasm of others to deliver something, start with the basics, where is the value to the business?
High initial costs
Another genuine concern. It is not just high initial costs that worries the senior leadership teams, but also spiralling costs. Making sure the business case includes the peripheral costings will ensure that the financial picture is realistic.
The CFO will see an increase in license cost for an AI & Automation project, that is pretty much a given. A strong Business Value Assessment is required to show the business where the return will come from, and in a way that they will see reflected in meaningful data. This might be an increase in sales, lower cost of sale, faster lead to order or order to cash, and maybe even a reduction in FTE costs.
What is the scope for the project? Are you applying this new solution to the entire client base, are you deploying it to the whole user base? Do you need to do it all this year? Or are you better off phasing it? There is a lot of ‘I want’ from users but that doesn’t mean it is the right place to start.
The other thing that needs to be built into your numbers is the peripheral costs. Including licensing uplift to deploy is one thing, but if there is an underlying £500k project to secure and clean up your data, then this will dramatically change the appeal of a new solution. Be thorough and open about the true cost to the business.
Done properly, most AI & Automation projects will see huge and rapid returns, but not if you start off from a flawed foundation.
Data Privacy and Security
This, in real terms, is nothing new. Any privacy or security gaps that will be exposed, already exist. There is no dodging this. AI will not introduce new privacy and security holes, but it sure as heck will surface any a lot faster than ever before.
Ensure that your current posture is as good as it can be. Data lifecycle management through the utilisation of Purview, Advanced SharePoint Management, classification and labelling etc. Also, any ROT (Redundant, Obsolete and Trivial) data will need to be removed out of scope as this directly impacts the accuracy of the responses your solution will provide. This is especially important when providing information to your clients who could hold you legally accountable for any misinformation provided. Court rulings already align with the opinion that any AI services are equally as accountable as an employee. ‘The AI did it’ is not a justification.
Regulatory and Compliance Concerns
Navigating the evolving regulatory landscape is crucial. Non-compliance can lead to legal issues and fines. The thing that makes this more interesting is that the EU AI Act has been released and whilst vague in places, it is very clear that the ‘Unacceptable Risk” category will come into force in Feb 2025, that is not far away.
There is much of this that is unclear, but the ‘unacceptable risk’ falls largely into the bracket of common sense and aligns to policies that most organisations will already have in place. No hate, racism, sexism or unethical activity. The regulations are set to become more refined between now and 2028, by which time the Act is expected to be dealing with the low risk use of AI. In short, the EU AI Act is tackling the severe cases first, whereas most businesses will be adopting the lower risk use cases first.
There are tools out there to help with AI regulation, but these, like the use of AI, is still in its early life. Adhering to the EU AI Act regulations today will help you develop your strategy from the beginning instead of having to retrofit regulations at a later stage.
Conclusion
All of these things are addressable. There is no need to panic, but amongst the noise and pressure to deliver an AI solution, don’t get pressurised into making rash decisions that you will rue later. Instead, review where your focus for AI is, consider it properly, and hone in on the first few manageable projects that will showcase the true power of AI within your business.
コメント