There’s much to discuss, debate, and disagree about in technology today. Evangelists and skeptics alike hold strong views and are not shy about expressing them. And not to mention us media folk, of course. We can’t recall any other technology occupying conversational space the way AI has. Where, and how, would one find ‘reality’ in the spectrum spanning “it’s all hype” on one side, and “human beings will be redundant” on the other? Well, we do the obvious thing and look demonstrable implementations and see just how far the best of them have come.
In late January this year, the World Economic Forum released the second edition of what they call MINDS. The acronym, Meaningful, Intelligent, Novel, Deployable Solutions, may seem like it is trying too hard, and be that as it may, the report and the case studies offer us a look at specific AI successes, and the aggregated learnings from respondents (those who submitted their case studies). The report, produced by the WEF in collaboration with Accenture makes for an interesting dissection and reflection with 30 featured case studies. So here’s a discussion on some standout findings.
Snowballing impact?
Roughly 75% of MINDS applicants reported reinvesting returns from current AI projects to fund expansion into new functions, while maintaining separate budgets for longer-horizon bets.
If this figure is something to go by, these respondents, and by extension, a significant portion of large enterprises implementing AI in some area, are not looking at the technology as a point solution only. It’s a strategic investment where accrued savings enable deeper entrenchment of the technology. A short-term peak in the evolution cycle is yet to come.
Can’t DIY always
We knew this instinctively, but the report reinforces it. About 30% of the respondents stated their projects were developed in partnership with, or utilized the expertise of, specialist technology providers.
As the AI landscape gets more crowded before interoperability (and consolidation?) comes in, building in-house is likely to become even more challenging. Specialist providers should see takers for their expertise.
Gen is passé, it’s agentic now
Summarize a report, correct a mail draft for language, generate an icon to use in a presentation, tell me a joke — this is how most individuals or employees use AI (although the jokes are a disappointment). The real game is in how organizations use agentic AI. Next-generation, workflow-spanning AI systems have crossed into mainstream practice among leading adopters. That’s not us saying it; that’s over one-third of respondents saying they have integrated agentic AI into their current solutions.
Replacing or co-working with humans?
This one reads like a bit of a relief, with one of the five themes in the report headlined “Amplifying strengths when humans and AI work together within a changing workforce.” There is no number that tells us what percentage of respondents said so. The skeptics amongst us would go, “How convenient.” Understandable. But taking this at face value, and applying our judgment, it stands to reason that “successful AI adoption starts with people and organizations redesigning work to augment human expertise with AI” (quoting from the report). Of course, that’s not to dismiss the issue of job loss, and that’s a wholly different discussion for another article.
The biggest learning here is that AI, more so than any other transformative technology, should not merely be a boardroom mandate, but should be co-created by the employees who are closest to the processes, enabling them to be more efficient. One of the featured case studies exemplifies this: Sanofi, working with OAO, enabled 60,000 employees to co-create over 1,300 AI use cases, reporting a 10x ROI.
Talent scarcity, anyone?
When juxtaposed with the fear of job losses it’s easy to overlook the immense potential for technology, even non-AI, traditional, rule-based computing, to alleviate opportunity cost for businesses, and even sheer human cost in some cases (healthcare is a striking example), arising out of talent scarcity. But this needs to be talked about. Perhaps the best win-win use case of AI is in areas such as high-tech, and, contrastingly, areas with numerous human touchpoints, where lack of human expertise or availability leaves profits or betterment on the table.
The MINDS report has two such examples. AMD and Synopsys deployed “reinforcement learning” and agentic workflows to handle more of the execution burden in semiconductor chip design, doubling design speed at one-fifth the cost where deployed, without displacing the expert engineers whose judgment remained essential.
In Chinese healthcare, Landing Med’s AI-enabled cervical cancer screening system has achieved 6% more correct screens in the country’s remote provinces, not by replacing pathologists, but by making them five times more productive.
Responsible AI. Let’s have some (more) of that, please
One of the five insights summarized in the report calls out responsible AI as a requisite for scaling, and not merely as a checkbox governance item. AI guardrails are something even the most vocal supporters of the technology call for. While the report does not mention examples from its respondents with regard to AI governance, it does explicitly state, “Confident AI adoption requires trustworthy systems, prompting organizations to embed technical controls and right-size human oversight for automated decision-making.”
The specific call for a human in the loop when it comes to governance should address (if implemented in letter and spirit) many of the detractors’ “AI gone rogue” scenarios.
To borrow and modify a line often used in spy thrillers, the guardrails have to work every time, but the AI has to go wrong only once (and in an inopportune scenario). Businesses and individuals alike have to know what their AI is doing, and if something doesn’t feel right or smell right, then pause and find out.
There’s a lot to unpack and analyze in the MINDS report, and there are other state-of-the-AI reports, too. We picked this one because it curates the most demonstrable use cases from across the world and industries, and the learnings from them. You can access the full report and summary on the MINDS page here: https://initiatives.weforum.org/minds/selected-minds.
[AI tools were judiciously utilized for research and summarization in preparation of this article. Analysis and writing are human.]
