To take the pulse of technology and business trends, Dell Technologies regularly surveys 800 IT decision-makers across six countries (U.S., U.K., France, Germany, Brazil, and China).
This month, we cast a spotlight on AI and asked IT decision-makers what their three-to-five-year forecasts are for AI. The results are now in. Here are the findings from Brons Larson, AI strategy lead for the Office of the CTO, Dell Technologies.
We’ve just asked IT decision-makers about how AI will revolutionize business in three-to-five years. What strikes you the most?
That AI is on the cusp of upending the way we do business—according to IT decision-makers around the world.
For starters, they expect it will fundamentally change their role: Nearly three-quarters (71 percent) believe AI will impact most or all of what they do in three-to-five years. While nearly four in 10 (38 percent) say that AI is already part of every interaction they have with technology. A further 37 percent believe this will be the case in one-to-two years.
What’s more, nearly half testify that their business is already putting AI to good use, with 48 percent saying their organization has accelerated data-driven insights with next-generation algorithms, and an additional 36 percent presaging this will happen in one-to-two years.
In fact, many expect AI will open the floodgates to new, superlative opportunities: 38 percent believe AI has already allowed their organization to surpass human limitations. A further 36 percent believe this will be possible in the near-term. However, they’re probably thinking about “surpassing human limitations” in quite pedestrian terms such as AI being able to crunch numbers and find patterns quicker than humans. I doubt they’re thinking about artificial general intelligence (AGI), which is still a way off, or artificial superintelligence (ASI), which remains the stuff of science fiction.
They’re equally optimistic about AI’s potential beyond their four walls: 87 percent predict autonomous vehicles will be a reality in three-to-five years, thanks to AI constantly learning the rules of the road and knowing how to respond to unpredictable human behavior. This timeframe is unrealistic given the many different hurdles that need to be overcome first nonetheless, AI has still made autonomous cars a distinct possibility. Meanwhile, 84 percent foresee manufactured goods being assembled by AI-powered, data-driven robots that make fewer mistakes.
Cumulatively the research points to the perception that businesses will strive to solve some of their most pressing problems with AI.
Are respondents right to believe AI will transform business and even the world as we know it?
AI comes with huge promise, but the current wave of AI, based on statistical learning, also has its limitations. As discussed here, explainability is an issue, as are high error rates.
So, while respondents are right to hand-off some tasks to machines—71 percent are doing so now or expect to do so in one-to-two years because machines don’t get tired, stressed, or sick—we also need to put human checks and balances in place. After all, stochastic-based intelligent machines can still make mistakes and when they do, their mistakes can be very hard to unpick, explain, and rectify so that they don’t happen again.
Why? Because neural nets effectively operate as black boxes and the data that they’re fed is often imperfect.
Today, 83 percent opine machines will diagnose disease more accurately than human doctors. There is certainly a strong case for using AI in healthcare, but I would argue we’re still in the era of human-machine partnerships and that rigorous human oversight is essential.
What does the research tell us about how businesses can build trust in AI?
As long as AI continues to operate as a black box, the technology will struggle to earn users’ trust. Some businesses are awake to this challenge and are taking steps to provide more transparency.
Just over half (52 percent) believe their organization can ensure data traceability and expose bias, as well as ways to remedy bad bias. That’s a bold and encouraging statement. Although it’s important to interrogate how they are doing this.
Some businesses are tackling the trust issue through lobbying for better regulation or education. Just over half (55 percent) said they would create educational programs for employees and customers to outline how they use AI technologies. And 45 percent would recruit AI explainers.
What does the survey tell us about the future of AI?
AI remains the posterchild of the IT industry—the study bears this out. Of course, AI’s been around for many years, but the move to deep learning has reawakened its potential. However, the most exciting development is yet to come—when Third Wave AI becomes mainstream.
Whereas Second Wave AI is based on statistical learning, Third Wave takes a model-based approach that requires virtually no training and understands context and nuance.
Once Third Wave AI supersedes its earlier incumbents, the opportunities to delegate to AI will increase immensely. By this point, we should already have classified which tasks and skills should remain the preserve of humans. Some of this thinking is already underway, with 41 percent apparently working with employees to develop human-specific skills that complement AI but can’t be replicated by a machine.
In fact, the Institute for the Future urged businesses to explore this back in 2017, when it produced a report recommending businesses create a framework in which human strengths are programmed and integrated into machine intelligence, paving the way for humans to start offloading tasks that are better suited for machines. It noted that “AI explicitly designed for collaboration will help build capacity in machines to improve their understanding of humans.” Third Wave AI would realize this near-seamless exchange between people and technology.