A pathway to AI Adoption for the Irish public sector
AI is the topic on everyone’s lips today, writes John Stobie, Regional Vice President of Public Sector Sales for Salesforce Ireland.
As I sit down to work on this piece, I am listening to Patricia Scanlan, the Chair of the AI Advisory Council in Ireland, sitting for an interview on RTÉ’s Morning Ireland about the many challenges and benefits of AI. A day seldom goes by without a news story about the evolution of this technology. Generative AI and its rapid ascension to the top of the technology stack has thrown a spotlight on a capability that Salesforce customers have been using, for over a decade.
In this article, we focus on the fact that AI is not an end itself, but that used correctly it can energise transformation and innovation within the public sector. AI is an accelerator which can personalise services and engagement, increase the productivity of government staff, and drive impact by allowing the targeting of limited resources to deliver the best outcomes. Governments face ever-growing pressure to improve the efficiency and personalisation of their services and overall interactions with citizens. According to a study by the Boston Consulting Group and Salesforce, “93 per cent of respondents expect government’s service quality to be on par with leading tech companies, private sector and digital governments”. We will discuss AI under a number of headings:
- If AI is the engine, data is the fuel;
- In the flow of work; and
- Trust is paramount.
Ultimately, we must understand that AI is not a silver bullet and to get the benefits of this technology, like any, we need to walk before we run.
If AI is the engine, data is the fuel
It is important to recognise the symbiotic relationship between data and AI. Ultimately, it is not enough to just amass lots of data, it must also be connected and structured. Salesforce’s citizen relationship management (CRM) is built with the citizen at its heart, and information about that individual wraps around it. This includes key attributes, benefit history, touch points, and other key data elements that not only provide agency employees with a rich view of their customers, but also serve as inputs to AI models.
“Ultimately, we must understand that AI is not a silver bullet and to get the benefits of this technology, like any, we need to walk before we run.”
John Stobie, Regional Vice President of Public Sector Sales, Salesforce Ireland.
However, not all data about the citizen sits in the CRM. Within any agency, dozens of other systems contain citizen data, including ERPs, payment systems, document repositories, and more. To join up this broader set of data elements, Salesforce’s Data Cloud enables agencies to connect and haronise data from other systems in real-time. With powerful identity resolution and graphical mapping capabilities, Data Cloud delivers the essential foundation to both traditional and AI-enabled citizen services. This complete 360° view of the citizen supports improved interactions and also better utilisation by AI models.
Just recently, one of our global postal customers revolutionised their digital marketing and customer experiences with AI and data-driven communications. Salesforce Data Cloud has been transformational to the business, taking over 11.5 million contacts and turning that into 3.3 million unified profiles, with rich engagements and transactional data, within six working days.
In the flow of work
To derive the greatest value from AI, it should act as an enabler within a process, rather than as its own functional area. When embedded into existing systems and business workflows, it ensures easy access for employees and customers alike, and minimises the implementation and maintenance work required from IT. It is important to not only feed an AI model a training set and prompt, but also the broader context for that prompt. This is known as ‘grounding’, where the surrounding information about the work item and the customer are fed to the model alongside the prompt.
Equally, the result should be returned in a way that is immediately useful to the agent. For instance, a draft email to a citizen should show up in the drafting screen itself, so the agent can easily make edits before sending. By embedding the technology, it also allows for seamless and automatic upgrades, including new technologies and capabilities. This has always been the Salesforce approach to innovation, delivering three annual functional releases for nearly a quarter of a century.
Trust is paramount
Government must handle citizen data with the utmost care and ensure services are delivered with fairness and precision. This requires an understanding of the risks associated with AI and how to control them. Salesforce have outlined five trusted AI principles to guide both our own development and our customers’ use of these technologies. The use of AI must be: responsible, accountable, transparent, empowering, and inclusive.
Salesforce’s Trust Layer helps to mitigate these risks, including protecting personal and sensitive information by not feeding it into the model, controlling for toxicity and hallucinations on the return, including with a human in the loop to review the output before it goes to the citizen, and memorialising the use of any AI-generated output. These are minimum-but-not-sufficient measures; however, any commercial AI solution should incorporate such controls. Experimentation to acquire institutional knowledge and understand the nature of such risks is crucial to safe and successful deployment at scale.
AI is accessible wherever you are on your transformation journey
AI continues to evolve rapidly. Customers should look to the old mantra, “think big, start small, scale fast”. Wherever you are on your AI journey, it is essential to understand the maturity of your organisation for the adoption of this technology. To help you get going, talk with us about our AI vision and capability assessment tool (VCAT).
T: 0800 086 8949
E: jstobie@salesforce.com
W: www.salesforce.com
