PwC: The path to AI adoption is slow and steady

Executives are facing an AI 'reality check' as they reevaluate their approach for 2020.
15 January 2020 | 25 Shares

Time to take a step back? Source: Shutterstock

Specialists in AI represent the fastest-growing job role in the United States over the last four years according to LinkedIn while in the UK, AI companies secured $1 billion in the first six months of the year. 

Stats like these lead you to believe that the world of business is well underway with the adoption of AI and the revolutionary potential the technology carries across industries. At the same time, when companies with nothing more than a chatbot claim that AI is central to their product, shadows of doubt are cast on that perception of progress. 

A new PwC report adds some clarity. According to 2020 AI Predictions, implementing AI proper is proving to be more of a challenge than expected. While 90 percent of executives are eager optimistic about AI’s potential, many business leaders are facing a “reality check.” 

While last year, the same report found that 20 percent of executives planned to deploy AI “enterprise-wide” in 2019, that outlook for the next 12 months has been significantly chiseled down to just 4 percent.

Fewer companies will pursue AI at scale in 2020. Source: PwC AI Predictions Survey

“Corporate America is — rightfully — still focused on capturing the expected US$16 trillion in AI gains in the next decade,” the report states. But for the next year at least, investments will be focused on investigating uses (42 percent) and pilots within discrete areas of the business (23 percent). 

We may then be able to expect organizations to market themselves less on their uses of AI as, behind the scenes, they seek to develop products and services where AI is a true differentiator with real disruptive potential for their industries and markets. 

In its report, PwC provided a ‘blueprint’, featuring priorities for businesses to follow in order to acheive a more transformative payoff in the years ahead.  

# 1 | Embrace the boring 

Not all investments in AI will serve as shiny new assets to enhance customer-facing products. One of the technology’s key opportunities will be in streamlining in-house tasks which will enhance efficiency across operations, free up staff time and save money. 

Bringing AI into the background, where menial tasks can be a burden, is a solid starting point for deployment. Meanwhile, in the face of a cybersecurity staff shortage and greater awareness around data privacy, AI can be boon to cyber defenses and fraud prevention

# 2 | Don’t stall on training

Offering training programs surrounding new technologies, AI included, can be an effective strategy for attracting new talent and retaining existing staff, but it shouldn’t stop there.

Both companies are employees alike by ensuring the skills that have been practiced in training are put to practical use, allowing them to continually enhance those skills in a setting where they will become increasingly necessary. 

The report also suggests teams should become “multi-lingual” in tech and non-tech skills so that colleagues can collaborate on AI-related challenges and better discern which problems AI is most suited to solve. 

# 3 | Address the risks 

Concerns around AI’s impact on job and skills displacement continue to dwell. While the oft-touted (and justifiable) line is that the technology can help free up time for higher-value tasks, organizations should also prioritize laying down foundations for maintaining strong ethics. 

This may comprise collaboration with regulators, customers, and across the team internally while establishing an AI ethics board— featuring technical and non-technical members from across all departments— to address issues, concerns and risks as they arise will serve organizations well for the longer term. 

# 4 | AI needs continual learning

Unlike software development, AI models require a constant ‘test and learn’ approach, where algorithms are continually learning and data is being refined. As much high-quality data as possible as key to the pace and level of development. 

The most effective way a business can develop AI models is by running them 24/7 across broad and multiple operational systems, alongside wide automation and data analytics initiatives.

# 5 | Part of a broad toolset

A key point to remember around AI is that it must serve in unison with a broad sweep of other technologies, solutions, and operations, it’s not a standalone solution, but part of a move towards automation or as a component in wider business strategy.

Businesses are wise to pedal back on plans for deployment; commitment to training and understanding ROI should come first, while other automation technologies, such as RPA, can be an equally worthy entry point.