The artificial promise: 5 reasons Gen AI projects fail and how to get yours right the first time.

There’s no denying that 2023 was AI’s breakout year. But is that big bang now fading with a whimper?

Despite initial excitement, 80% of AI projects have failed this is down to rushed implementation, unclear strategies, or the belief that such solutions can simply be turned on and work flawlessly without any planning. Unstructured approaches like this tend to deliver disappointing outcomes.

But AI isn’t going anywhere; 79% of employees report having used generative AI, and 22% use it regularly at work, with 40% of organisations plan to invest more in generative AI. This proves just how apparent the appetite for AI is, but the results are not being achieved. 88% of CEOs have an AI vision, but only 25% are seeing tangible returns, showing further evidence of a divide between theory and practice. Many companies are just starting to dip their toes into this new realm2 with 89% of businesses using generative AI in some processes, yet only 23% have fully deployed it, with most projects stuck in the developmental stages. In fact, you could categorise businesses through this lens:

• The leaders: These businesses have generative AI are the core of their business, proliferated across their teams and reaping the rewards. Interestingly, two thirds of this group are traditional incumbents meaning success is not limited to digital natives.
• The experimenters: These companies are stuck piloting generative AI solutions, meaning they cannot scale their initiatives and their investments and deployments lack cohesion, leaving valuable revenue streams on the table.
• The laggards: This group is currently taking no significant action to advance or implement generative AI. They typically lack the infrastructure, data quality, and/or leadership buy-in required to begin embarking on their generative AI journey.

Most businesses fall into the category of experimenters and laggards.

This begs the question: why are so many projects falling short?

Unearthing the roadblocks to AI success

There are several reasons AI projects are not achieving their intended results, including:

• Tech debt: the hidden barrier to victory
  Legacy platforms and systems don’t often communicate effectively, creating siloed data that AI can’t fully utilise, negatively impacting the quality of the output as the available datasets can be outdated, be contradictory to what another data source says, or no longer be relevant to the current business context.
• Data disaster: inconsistent data derails AI
  Inconsistent, fragmented and incorrect data undermine AI performance. The importance of clean, correct, high-quality data is essential to building a strategy based on all available information, not just some of it
• Regulations: compliance restrictions can halt progress
  Companies must adhere to regulations concerning the handling and processing of various data types, including how it’s consumed, generated and shared. This is also true regarding the development and distribution of AI. When looking at the EU AI act for example, which impacts UK businesses serving EU customers, your AI application must meet certain criteria and non-compliance is costly
• Rushing in: unclear goals lead to unclear results
  Many organisations have jumped into AI projects without a clear understanding of their purpose or strategy. Often businesses make investments into AI that are non-coherent, meaning AI applications aren't embedded and serving multiple functions, but rather acting as point solutions. It's also likely that without a clear goal, your workforce will not be properly enabled on how to use AI, damaging uptake and adoption, weakening AI investment. A lack of accurate data aligned with clear goals means a project might only be identified as a failure once the results are in, making for a costly AI experiment
• ROI misery: the difficulties in measuring results
  Where many AI benefits are tangible, it can be difficult to gauge the success of an AI project, especially when expectations aren’t clear from the outset. If challenges like those mentioned above aren’t considered and addressed, project outcomes are harder to define and, as a result, more difficult to measure. This requires both changes in approach and mindset

Reset expectations – how to make AI work for your business

AI has the potential to become a fundamental part of your company’s processes, but it needs to be managed correctly. Every company has unique goals, but the demand for well-prepared data remains the same, making it crucial for businesses to first modernise their infrastructure, policies and processes to prepare for AI.

With a successful and reliable data management process in place, integrating AI into your systems becomes more seamless.

Another way to ensure successful AI implementation is by setting clear and achievable goals, implementing a transparent process and ensuring the ROI can be demonstrated effectively. Staff training, summarising support tickets and simplifying technical documents are all AI use cases with clear outcomes. Supported by high-quality data, they can deliver measurable outcomes and meet expectations.

How to get started with generative AI for your business

AI is only as good as the input it is given, highlighting the need to first master how your business handles data. You need to harness the full potential of your data, ensuring it is clean, secure and ready for AI applications.

Claranet offers Data Platform services to build a robust foundation for data management, pulling in data from different sources and transforming it into high-quality, consistent information that supports better decision-making. This includes essential activities to support your data-driven transformation:

• Data Engineering: Improving and maintaining the quality of your data through pipeline management, orchestration, data integration, CI/CD, data modelling, and more.
• Infrastructure management: Ensuring the cloud services and infrastructure supporting your data initiatives such as storage, compute are optimised.
• Analytics support: Management of existing reports and dashboards including the monitoring and support of analytics models.

Your AI opportunity awaits

AI isn’t just about technology; it’s about using data as a strategic asset. However, to get the best results, that data must be prepared effectively. Businesses must reset their expectations and avoid diving into AI implementation without appropriate planning. By leveraging Claranet’s expertise in data platform services, you can bridge the gap between your vision for becoming an AI-powered organisation and the reality behind achieving those results.

Find out more about our Data Platform Services and how you can take your first steps to becoming truly AI-ready.