By Laith Al-Bazirgan, Digital Evangelist at Endava
Artificial intelligence (AI) has rapidly evolved to become a defining force in the modern zeitgeist. In just a few years, the questions being asked about AI in boardrooms have shifted from “Why should we adopt?” to “When should we adopt?” to “How do we adopt?” to “Why have we not already adopted?”. This latest attitude is one that seems to be especially pertinent to generative AI, the new tech sensation that has gripped the imagination and launched a thousand conversations about efficacy, integration, privacy, and security.
Chatting naturally with a generative AI tool like ChatGPT is a mesmerizing experience. And yet we seem to have forgotten about the established and proven variants of AI in the face of the generative generation. Remember that PwC prediction of a US$320-billion impact on the Middle East economy from AI by 2030? It predated ChatGPT, Google Bard, and Bing AI. We have been living the AI revolution for some time now, and for enterprises to make sure they share in that multibillion-dollar windfall, they must avoid the trap of adopting the latest thing simply because it is the latest thing. They must shake off the gen-AI hype and get back to business, asking things like “What is our mission?”, “How do we do it better?”, and “How do we pick the right flavor of AI to help us get there?” What follows are some use cases that should help focus the mind.
1. Deep learning
Sifting through what humans could never accomplish in a thousand years is quintessential AI. Deep learning (DL) is a type of machine learning that uses artificial neural networks to crunch numbers at scale. Indeed, given enough data, these algorithms outperform baseline machine learning. DL can be used for a range of analytics use cases based on its ability to extract complex features and entity relationships from data. It is good at conversational tasks (understanding and speech) as well as image and video processing, even saving scientists from hours of eyeballing specimens under a microscope by taking over and doing the task significantly more quickly. It is no exaggeration to say that this speeds up innovation, and not just in science. Any industry that places a premium on such R&D will benefit. Deep learning also brings value to customer service. It improves everything from the individualization of digital experiences to the parsing of documents.
2. Natural-language processing (NLP)
While large-language models, like ChatGPT & Co., fall into this category, NLP is so much more than LLMs. Many technologies oil the gears of understanding between computers and humans. Any speech-to-text tool uses NLP, but most predate the widely available LLMs of recent months. Virtual chatbots are another application. Virtual assistants have the capacity to enhance employee productivity. Effective semantic translation tools enable more effective collaboration in multicultural workplaces across the region. And the Middle East’s digital natives can get access to much-vaunted self-service experiences. NLP can also sift through tens of thousands of feedback forms, emails, online reviews, and social posts in search of patterns and insights; sentiment analysis gives brands and their marketing teams a huge leg up.
3. Predictive analytics
When statistics and ML team up, we get predictive analytics (PA). Meaningful patterns found in historical data are powerful indicators of what is to come. If they duplicate certain determining factors, decision makers will see the probability of the recurrence of previous results. They can take informed action, which allows them to capitalize on positive predictions or avoid the worst impacts of negative forecasts. You anticipate a demand spike, so you top up your inventory. You foresee a dip in energy needs, so you dial down your consumption. Healthcare organizations can personalize wellbeing plans for patients. Manufacturers and utilities can get ahead of equipment breakdowns. Insurance underwriters can fine-tune assessments.
4. Computer vision (CV)
This is a type of deep learning used to process images for various use cases. A neural network consumes a huge dataset of images, assimilates human guidance, and learns. Advanced CV models can spot safety hazards on live CCTV footage. Autonomous vehicles know the difference between a brick wall and another moving vehicle. And production lines can automate QA by allowing a CV module to look for defects and flag them. While all these examples can be fulfilled by a human, CV does it faster and more accurately.
The right stuff
AI adoption is on the rise and on course to fulfil PwC’s projections. As some wring their hands over job losses, we should note that, to date, AI is being widely adopted to improve employees’ experiences rather than replace them. According to McKinsey, as of May 2023, 62% of GCC organizations use AI in at least one business function, so adoption is well underway and shows no sign of slowing. Doom-and-gloom predictions of supplanted workforces, in Endava’s view, are overblown and fail to account for generational norms where younger, tech-savvy workers expect to be supported by advanced, relevant technology (not necessarily generative AI).
In all the use cases above, I show how human-centric AI can supercharge the journey to the fulfilment of business goals. Generative AI has many applications, and some of them will fit use cases in a range of Middle East businesses. But it would be a mistake to be so focused on this single segment that you miss the impactful opportunities that other forms of AI present.
