Machine-scale performance will transform the very nature of competition in the global marketplace.
For a few years now, intelligent automation has done great things for productivity and efficiency. We’ve measured its performance relative to how quickly and effectively humans can complete tasks.
But as humans, our mindset is anchored to what people can do in an hour, a day, a week; we measure performance within the boundaries of human biology and earth-delineated cycles, which is overly constrained.
In an AI-enabled world, humans are still very relevant, but human-scale metrics to measure performance are much less relevant.
AI sets the stage for a new paradigm of value based on what people and machines do together. This machine-scale performance will transform how companies are structured, how they invest and the very nature of competition in the global marketplace.
The 10 million percent efficiency increase
A profound example of how AI can accelerate work has come from the world of astrophysics recently.
Normally, astrophysicists review images from the Hubble Space Telescope and collaborate to determine if a smudge on an image is a new gravitational lens, a finding that impacts math on a galactic scale. This is a cumbersome, manual process that takes months. As a result, in the last 50 years, there has not been much progress.
But after just eight hours of training an AI engine, it learned to do the work of a team of highly trained astrophysicists, the equivalent of months of manual effort, and to do it in one third of a second. That’s an acceleration of 10 million percent.
And now we are seeing the benefits: lifetimes of scientific work performed every week. In the past year, astrophysicists have discovered a 10th planet in our solar system, the largest black hole, the furthest star in the known galaxy and more inhabitable planets. Our understanding of the universe has been materially impacted by one AI model. Just imagine what AI could do for an enterprise.
Beyond efficiency, beyond human
Of course, companies cannot expect AI to deliver efficiencies by that order of magnitude every time, nor is it necessary — the important thing is that AI can, in the right circumstances, deliver operational performance improvements that cannot be measured in strictly human terms. We need to think bigger when calculating the business value of AI.
For example, in HR one of the biggest burdens is the sheer volume of candidate applications to review and interviews to conduct — the process of sifting through a million applications and finally making offers is very labor intensive and inefficient. It is also believed to require a human touch. AI can replace many of the tasks in this workflow, however, and perform as well or better than humans in almost every way.
AI can be trained to read resumes and assess quality and experience; AI can correspond with candidates through email and gather necessary information to complete forms; AI can even conduct interviews and apply artificial empathy to evaluate nonverbal responses, such as body language and posture. AI is being applied successfully to automate almost the entire recruitment process, creating a fundamentally new model of efficiency for HR.
In finance, detecting fraud in varying contexts from vendor invoicing to employee expenses is normally a matter of human-led pattern spotting, programmed algorithms and even luck. Analyses are also based on data samples and extrapolation. But with AI, we can do more than look for known patterns, we look beyond the historical factors and rely on machines reviewing 100% of the transactions, learning (without human supervision) from actual outcomes to improve its pattern spotting abilities over time, optimizing performance. We can also see, read about and generally observe human behavior at scale in ways that would be entirely impractical if performed by humans, such as analyzing every workplace text or email (subject to privacy regulations) in real time.
To fully harness the business value of AI, we need to look beyond traditional, human-scale KPIs and consider the potential for machines, working without constraints, augmenting humans in the total performance of the enterprise.
Of course, this is all academic without a world-class technology and data infrastructure. AI is not implemented, but rather it is applied, and it can only be applied as a capability on top of existing digital capabilities, including an advanced platform like SAP Leonardo, combined with the skills needed to operationalize and sustain it.
Together, though, we can welcome the newest vector for global competition: machine-scale performance powered by SAP.