So many businesses I talk to these days are exploring the exciting yet complex world of AI. They are asking themselves, “How do I leverage this technology?” According to a recent Forbes report, research reveals a significant gap in AI training perceptions. While nearly three-quarters (72%) of employers say their employees are at least adequately trained on AI, only just over half (53%) of employees share this view. Moreover, nearly half (45%) of employers cite unprepared data as a barrier to AI implementation.
Key Takeaways
- Businesses face a gap in AI training perceptions, with employers overestimating employee readiness for AI.
- Many organizations struggle with siloed data, making it hard to leverage analytics to work effectively.
- To succeed in AI implementation, agility is crucial; adapt strategies to rapidly evolving technologies and regulations.
- Starting with concrete use cases can help businesses create manageable ‘data ponds’ before developing larger data lakes.
- As AI progresses, organizations must prioritize data privacy, security, and ethical practices to mitigate associated risks.
As Forbes notes, this often stems from data being siloed across the organization and being inaccessible or ready for AI. This can severely impact how analytics work and overall AI effectiveness.
Which businesses are best placed to benefit from AI?
People often ask me which sectors are leading the way in leveraging AI. I would argue the key differentiator between leaders and laggards isn’t necessarily the industry. It’s how mature the organizations are in their data transformation journeys. Some organizations may not need to contend with legacy systems. It’s easier for them to be agile in that case. Some of the businesses may have significant historical data. They could also be generating new data daily. In such cases, there could be a greater opportunity to tap into the amassed data.
According to research company Statista, “the total amount of data created, captured, copied, and consumed globally” is forecast to more than double, from 149 zettabytes last year to more than 394 zettabytes in 2028. Organizations usually own data from various sources and collect it in multiple formats (structured and unstructured). Such data needs to be carefully processed before it can become a valuable asset for the organization. This is very often the biggest challenge.
AI implementation – the need to be agile
AI technology is dynamic and fast-changing, with innovations emerging over months rather than years. This means that today, as a business leader looking to quickly and effectively exploit AI’s full potential, you must be agile. Yes, you need a strategy, a north star that provides an ultimate guide to your ideal outcomes. Meanwhile, rapidly adapt to technical innovations and regulations.
The most valuable support we can offer our clients is helping them build an ecosystem that fosters agility and flexibility. This allows them to be ready to adapt to new developments while continuing to effectively deliver based on today’s needs.
Getting your data strategy and infrastructure right
This means that getting your data strategy and infrastructure right is essential. As they say, “Garbage in garbage out.” Data intelligence is vital to drive agility, resilience, innovation, and rapid growth in a digital-first world.
The main obstacle to integrating data and analytics into operations is the existence of silos. Many companies have spent years and millions of dollars trying to get their systems to share data and to speak to each other – and they’ve often failed. Data lakes are usually seen as the key to this integration and sharing of data. However, they can be slow, expensive, and cumbersome to build. Often, businesses are not sure what to do with them once they’re complete. I take a different approach.
Instead of starting from scratch and building data lakes from the ground up, I suggest starting with a use case. You can then work backward from these use cases to create “data ponds.” These smaller collections of data can address a particular need. This allows the business to become more agile. It joins up these data ponds to build a data lake at a later stage. Meanwhile, its ponds deliver value to specific departments. They provide a return on investment more quickly, delivering value to the business as you go along.
When it comes to implementing how AI and analytics work, businesses have great opportunities to learn from each other.
Getting more value from how AI and analytics work toward solutions
Finally, many business leaders ask me, “How can I extract more value from my investments in analytics and AI solutions?” Sometimes, AI is not even the answer. In some cases, an old-fashioned spreadsheet might be more appropriate for a particular task. In contrast, in others, a complex orchestration of Agentic AI models is required to achieve the desired business outcome.
Again, the secret is to start with the end objective in mind and ask, “What kind of value do you want to drive for your business?” Then work backward to derive use cases where analytics and AI interventions will be effective. We at WNS Analytics have developed sophisticated frameworks to prioritize use cases. We create an agile roadmap for AI and analytics interventions.
Keeping ahead of AI regulations
Governments and regulators are grappling with the rapid pace of AI innovation. While there is broad consensus on the need for regulations to benefit both businesses and individuals, companies that fall behind in adapting to evolving technology and regulations risk fines, disruption, and reputational damage.
The European Union Artificial Intelligence Act (AI Act), described by the European Commission as “the world’s first comprehensive regulation on artificial intelligence,” came into force. Meanwhile, regulations in the US and APAC are also rapidly moving.
With rapid developments in the AI landscape, regulations around AI will inevitably continue to increase. Beyond simply remaining compliant, being ready for the introduction of new regulations will be increasingly crucial for all businesses. We help clients with AI-driven solutions that align with relevant regulations, allowing them to focus on making AI and analytics work for their businesses.
The risks posed by AI
However, rapid AI evolution carries other risks. Privacy and security concerns dominate. AI and analytics often sweep up personal information, including sensitive data, to train the Large Language Models (LLMs) that perform tasks and answer queries. Cybercriminals can exploit some of this data to target individuals and organizations, while identifiable personal information may be made public or used without the owners’ consent.
To overcome these risks, organizations must anonymize and securely store all data. They should also obtain relevant consents and ensure that their systems and practices comply with regulations such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA). Additionally, they can develop ethical and security practices that go beyond regulatory requirements.
The generation of false or misleading information by AI, known as hallucinations, can pose serious risks for businesses. These inaccuracies can lead to poor decision-making and even regulatory consequences. One US law firm, for instance, was fined after relying on Generative AI to draft a court submission that turned out to have been almost entirely fictional.
Therefore, organizations must use specific, relevant, and high-quality data to train their LLMs. “The more the merrier” does not apply to data for AI training. Targeted AI solutions designed for specific tasks generally exhibit fewer hallucinations than their general-purpose counterparts. Human input is also very important when it comes to sense-checking and cross-referencing results produced by AI.
As AI and analytics work to push technology at an astonishing pace, organizations are only beginning to grasp the opportunities the technology offers. However, we can confidently say that identifying the right use case and selecting the appropriate technology are crucial to success.
