News & Analysis

GenAI Revolution? Not Yet, say Surveys

While consultants would have you believe that Gen AI adoption is happening thick and fast, the reality, according to recent surveys, is that implementation is still a far cry

Search the internet for generative AI and the first couple of pages displaying results could have you believe that an artificial intelligence revolution was happening right outside the window. But, go down to pages five and beyond, and the truth will out – that Gen AI hasn’t started changing the world and the way it works yet.  

As for vendors and solution providers seeking the attention of prospective enterprise clients by listing out its potential benefits, we’d urge you to save your breath. Because, recent studies are clearly indicating that while Gen AI has made it to most boardrooms, the response from within is far more nuanced that its proponents will have us believe. 

The major challenge in adoption appears to be the journey from a proof of concept to a working product. The technical complexities around implementation isn’t as simple as it was made out to be – whether the blame is apportioned to legacy tech stacks or the absence of people possessing the right skill sets. 

The main challenges are… we’ve always known them! 

A Gartner report notes that the top two challenges to AI implementation were finding ways to estimate and demonstrate value (around 49%) and the obvious lack of talent (42%). It says the “primary obstacle to AI adoption is the difficulty in estimating and demonstrating the value of AI projects. This issue surpasses other barriers such as talent shortages, technical difficulties, data-related problems, lack of business alignment and trust in AI.

And in case that isn’t convincing enough, another study by Lucidworks – an enterprise search technology company – shares that only one in four surveyed had successfully implemented a Gen AI project. Further, it perceives a spending slowdown with only 63% planning to hike AI investment over the next 12 months, as against 93% in 2023. 

Mike Sinoway, CEO of Lucidworks says, “the initial wave of enthusiasm for generative AI is being met with a more strategic approach. Businesses are recognizing the potential, but also the risks and costs. This study provides the insights you need to make informed decisions and chart a successful path forward with AI.”

And, that’s not all. A recent article published by TechCrunch quoted Aamer Baig, senior partner at McKinsey and Company as saying that their survey found just 10% of companies implementing AI projects at scale. And out of this bunch, only 15% saw a positive impact on earnings, suggesting that the hype has far overshot the reality. 

What exactly is going wrong? We knew that too! 

So, what exactly is going wrong? Respondents to various studies suggest that the slowdown in AI adoption could be the complexity of the process itself. A simple project with a handful of tech elements requires the right foundation models, the right quality of data, foolproof security and a broad reskilling of employees around areas like prompt engineering and implementing IP controls, where required. 

Of course, the original challenges resulting from legacy technology issues persist to this day with Baig noting that a major obstacle to achieving Gen AI at scale came from too many technology platforms. The question often is around the quantum of technology debt that a company carries and what sort of deficits need to be tackled. 

Close to two-fifth of the respondents to the Gartner survey pointed to a deficit of quality data as the top barrier to implementing AI solutions. And experts argue that it’s not the quantity of data at all. In fact, many say that companies must focus on limited data sets while attempting to create a Gen AI use case. 

What’s the way forward? Cut the clutter, for sure!

Data readiness was described as crucial to the successful execution of Gen AI and it would help if those piloting the project actually focus on data that helps with multiple use cases. Not only does this enhance the value of the exercise, it is also likely to get far more buy-ins from divisions within an enterprise.

Most of the experts also believe that the next six to eight months would be a test phase to figure out how much of Gen AI is hype, if at all. The proof of concept stage is past and now these tools need to generate direct benefits that are measurable by way of ROI. The need of the hour is to move cautiously on this journey, always seeking to find ways to measure the right metrics that justifies the increased costs.  

The respondents also opined that in spite of doing everything right, there is no guarantee that the AI initiative will generate the right results. Therefore, teams would be better served to avoid taking on too much and focus on the reuse options when and where these present themselves during the journey. Because reuse translates into delivery speed, which then keeps the businesses happy and invested.