Splunk: How to Avoid AI Implementation Challenges

While AI promises efficiency and innovation, Splunk warns that up to 80% of AI projects may fail by 2025 if not implemented strategically.

The potential pitfalls of AI adoption stem from various factors, including misalignment between technical efforts and business objectives, lack of clear goals and inadequate data practices.

However, with proper leadership, data management and a focus on achievable outcomes, businesses can harness the power of AI to drive meaningful results.

To find out more, James Hodge, Chief Strategy Advisor at Splunk, explains the key considerations for successful AI implementation and the future of agentic AI.

What are the key factors contributing to the potential 80% failure rate of AI projects in 2025?

I’m very positive about AI.

It is without a doubt, one of the most powerful and useful tools available today, in many circumstances, for scaling and automating.

Applied within the correct frameworks that are aligned to clear objectives it can deliver the right results.

The real crux of the matter for me, and where the risk of failure comes in, is that many AI projects are launched without well-defined goals, resulting in misalignment between technical efforts and the specific business problems AI is meant to address. 

If we unpack this further, there can be a temptation, and a certain pressure, for organisations to implement AI as ‘the latest thing’, and as an end in its own right, without absolute clarity about what they want the technology to deliver.

Companies that are seeing success are often starting small, understanding how the technology aligns with existing business processes, targeting specific business pain points and ensuring their data policies are keeping pace with their innovations.

They learn as they go and scale accordingly.

How can businesses strengthen their data practices and leadership to achieve realistic, high-impact AI outcomes?

Businesses must strengthen their data practices and leadership.

Centralising and ensuring data is clean and accessible lays the groundwork for efficient AI models that can quickly adapt and deliver measurable results.

Maintaining data integrity and complying with privacy regulations ensures that AI applications are both trustworthy and effective, particularly in industries like finance, where real-time data can improve fraud detection.

Also effective AI leadership and collaboration are essential in ensuring use of AI results in impactful solutions.

AI leaders must align initiatives with business objectives, focusing on achievable, short-term outcomes. They also need to bring in the expertise of data scientists, business leaders and domain specialists. 

When this works, the outcomes are brilliant.

Take the Ersilia Open-Source Initiative as an example, which is using AI models for the important cause of accelerating drug discovery, helping scientists and researchers in low-resourced settings cure local diseases and minimise costs.

It recently partnered with Splunk to streamline data ingestion and monitoring across this process, scaling its AI model hub from 100 to 500 models, allowing its team to research new drug discovery more effectively.

What are the potential benefits and risks of agentic AI for businesses?

Agentic AI holds the promise of creating systems and chatbots that can make autonomous decisions and adapt in real-time.

However, we are still far from realising its full potential.

Like previous technological waves, businesses need to focus on learning from existing challenges before diving into more complex, self-directed systems.

Current AI efforts often struggle with foundational issues such as data inefficiency, bias and underperforming models, which need addressing before they can embrace the full capabilities of agentic AI.

As with past innovations, it's critical to build a strong base of understanding and practical applications before scaling up to autonomous AI systems. 

Here at Splunk we operate a domain-specific, human-in-the-loop model of AI, where AI is intended to augment scale and supercharge human efforts, not replace them.

This ensures that outcomes and processes, ultimately, remain within the human sphere of control and decision-making. In our industry, there are complex, real-world, or ‘high-level’ decisions that still require the understanding and context that only a human being, operating in the real world, can provide.

How should companies approach agentic AI implementation to avoid repeating past mistakes with new technologies?

Companies need to establish clear governance frameworks.

These should ensure transparency and accountability in decision-making processes, enabling businesses to closely monitor AI activities.

Ethical AI design is essential here, where fairness, transparency and privacy is incorporated from the start.

Additionally, companies should adopt an iterative deployment approach, starting with pilot projects to test AI's capabilities and identify potential issues before full-scale implementation.

These strategies can ensure that agentic AI evolves in a responsible and controlled manner.

How is Europe positioned to lead in AI sovereignty and what advantages does this offer to European businesses?

Sovereign AI refers to a nation's ability to build AI capabilities and seize AI's potential.

Public policy discussion has focused on AI governance with regulations like the EU AI Act, while the location of data is often brought up in the context of sovereign AI. 

For European businesses to lead in the global AI landscape, there are many other critical factors that come into play, including the development of an AI-skilled workforce, local language-trained models, a secure and connected AI infrastructure and enabling uptake of AI across nations’ businesses and citizens.

Sovereign AI should be seen as complementary to global cloud-delivered AI, rather than a zero-sum game.

On the infrastructure side, for example, we expect to see enterprise AI, local cloud and AI providers, and hyperscalers all play a role, depending on the use case. 

Explore the latest edition of Technology Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.

Discover all our upcoming events and secure your tickets today.

Technology Magazine is a BizClik brand