Automation technologies – including physical- and software-based robots, artificial intelligence, machine learning, and other types of “learning” tech – are expected to change the future of work in a massive way in the coming decades. McKinsey Institute recently projected that over half of “worker activity,” or job tasks, has the potential to be automated as early as 2035. This tech will increase global productivity growth by 0.8% to 1.4% annually, the consultancy said.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
For businesses, this represents tremendous savings as the cost of producing goods and delivering services goes down. And it’s not too soon for companies to take steps to take advantage of these technologies. Indeed, those that haven’t yet taken steps may be left behind – permanently, said Josh Sutton, global head of data and artificial intelligence at consultancy Publicis.Sapient.
“I do have [a] concern around whether [AI laggards] will be here a decade from now or not,” said Sutton in a phone conversation about artificial intelligence and IT last week.
He points to two reasons: Consultancies like Publicis are seeing a major shift in business success stories toward companies adopting a services-based approach rather than a product-based approach. In addition, the rate of change in industry is increasing and the cost of vendor lock-in is decreasing. “If you take those two things in parallel, what you’re seeing isn’t just a technology wave. You’re seeing a fundamental shift in how businesses and enterprises want to engage from a commercial point of view, and [if you’re] behind that shift, [it’s] going to be … very difficult to catch up,” he said.
Companies that wait for AI technology to develop into what Sutton calls “canned solutions” run the risk of forever playing catch-up – or failing completely — because the rate of change is so fast.
AI challenges: Education plan
He advises CIOs to map out a strategy for AI technologies. But they face a number of potential obstacles, he warned.
The first big challenge is for CIOs to educate themselves on the technology and discern between vendor hype and actual product capabilities. “There’s a tremendous amount of noise in the industry and there’s a very wide gap between what many product companies position themselves as being able to do versus the reality of what they can actually deliver upon,” he said.
Organizational expectations can also complicate a CIO’s efforts to set an AI technology strategy. Business leaders have a range of attitudes about AI, Sutton said, from those who believe it’s the answer to all of a business’ problems to those who are dead-set against investing in such an emerging technology.
“In only extremely few cases does the business leadership of a company have a well-grounded understanding of what’s possible vs. what’s not,” he said. IT leaders need to work to set appropriate expectations.
AI challenges: Single vendor vs. best-of-breed choice
Once CIOs and business leaders have a clear sense of what is truly possible with AI technology, they will face the question of whether to adopt a single vendor’s platform or to take a best-of-breed approach. Sutton advocates the latter.
“I believe the industry is too immature to commit to any single player. And I don’t think any single player provides you the full suite of capabilities you’d want across an enterprise yet,” he said.
Indeed, the decision will be a strategic bet for CIOs. Should they align themselves with a particular vendor in the hopes that its AI technology develops the capabilities that the IT organization needs? Or, should they put together a best-of-breed architecture, which is more complex than the single-vendor strategy but offers more flexibility?
“I have a strong bias that the services-based best-of-breed approach is a less risky road for companies to go down,” Sutton said. “But that’s not to say that some companies won’t make the right bet on the company that does get to that true enterprise-level platform first and standardize around them and have a nice clean, simple architecture as an advantage coming from that. But it’s a big bet to make.”
AI challenges: Data governance, culture shock
The final big challenge that Sutton sees relates to data strategy, because many AI use cases are dependent on data, whether from internal or external sources. Businesses need to have solid data governance policies around their internal data, and understanding how to augment that data with data from external sources is also very important.
“Even your data scientists and data teams that you historically have had on staff might not be particularly equipped to deal with the challenges that need to be addressed,” Sutton said.
That’s not to suggest that there’s a big shortage of people who are prepared to work with AI technology. Sutton said that he thinks there are a lot of technologists who are very interested in working with AI, and the tools are easy to work with. The skills required for AI development can be transferred from skills developed around other technologies, he said.
Sutton said he’s more worried about the cultural changes that businesses will face as AI technology is implemented. AI will increase human productivity, which will inevitably lead to staff reductions. Some of those reductions will be accomplished by attrition, but some of them will be more visible, much like those created by outsourcing in the past.
“Outsourcing was viewed with a very, very negative perception because it was taking jobs away and it created a backlash against the organizations and people within organizations driving that,” Sutton said. “I believe that there’s a risk that that might happen with relation to artificial intelligence solutions as well.”