By Sunil Senan, Senior Vice President and Business Head, Data and Analytics, Infosys
Globally, the adoption of artificial intelligence has been growing through the years. While early adopters and hyper scalers have been using machine learning and AI for decades, AI is now more readily available and embraced by organizations of all sizes with different levels of technology proficiency.
Businesses have been making use of classic rule-based artificial AI and simple automation for a while now. However, now, AI experts, and academics have achieved significant advances in AI, and it can derive value from massive amounts of data and develop novel consumer products.
Enterprise spending on AI is booming too. According to the research firm IDC, global spending on AI, including software, hardware, and services for AI-centric systems, will top $118 billion in 2022 and grow to more than $300 billion by 2026.
Most companies are trying to attempt their own advanced AI implementations to uncover hidden insights and construct new business models. A shipbuilding company, for example, uses an Al-based Condition Monitoring System on one of their ships to improve the operational availability and to reduce downtime of ship borne equipment. The system uses state of the art diagnostics and prognostics model to assess and predict the health of critical equipment.
While many companies have achieved basic AI capabilities, they now want to operate AI at
enterprise scale. However, a lot of AI investments aren’t paying dividends as they deal with
multifaceted challenges involving data processes, AI techniques, and interdisciplinary teams. Companies need to think differently about data and AI.
Infosys Knowledge Institute’s inaugural Data+AI Radar identifies why AI fails to deliver on heightened expectations. The survey found that most companies have not achieved advanced, top-tier capabilities. This, of course, leads to middling outcomes with limited satisfaction with data and AI results. More often than not, data and AI fail to deliver high satisfaction from users because most companies use only basic AI.
For a dipstick into what capabilities AI systems deliver, the Infosys Knowledge Institute developed the Sense, Understand, Respond, Evolve (SURE) taxonomy. The SURE framework includes four tiers, ranging from the basic sense capabilities (simple signal processing, such as being trained to recognize an object) to the most advanced, evolve level (a system that senses, finds causes, acts on recommendations, takes feedback, and refines its performance). Scoring survey responses across the framework revealed that only 15% of AI systems reach the evolve tier.
Extracting business value out of AI
The report also recommends three areas for improvement to generate real value from data and AI – develop data practices that encourage sharing, build trust in advanced AI systems, and focus AI teams on business goals.
Getting data right
Before putting any AI into production, companies must have accurate, organized data. To scale AI and unlock value, businesses need to focus on data-sharing capabilities and hub-and-spoke data management.
Data today is more like currency and gains value when it circulates. The study shows that the companies that shared data, in and out of their organization, are more likely to have higher revenue and use AI better. On the other hand, shortcomings in data verification, data practices, and data strategies continue to hold companies back.
Building trust
Strengthening ethics and bias management helps build trust and satisfaction in advanced AI. This grows increasingly critical as companies put more advanced AI systems to work.
Advanced AI requires trust in all directions – both in data management as well as in the AI models. Most organizations that are contended with their AI applications consistently have strong, responsible AI practices and are not just dependent on pristine data and perfectly programmed AI models.
A business-focussed AI team
Businesses also need to recruit multidisciplinary AI teams, including data scientists, experts in business problem, and senior executives. A diverse, business-focused team keeps Data+AI tethered to business priorities. It’s obvious that data scientists should be involved in AI work but experts in the business problem guide data scientists and AI systems in properly framing the problem at hand and senior executives are critical to scaling AI because of their knowledge of strategic operations.
—
Till now, advanced AI has eluded most businesses. However, scaling AI across the organization and unlocking business value will positively impact the bottom line and increase internal satisfaction with data and AI while adding up to $467 billion in potential profit growth, collectively.
