It’s not hard to find data that backs up a point of view. There are surveys suggesting that AI is taking jobs. Other polls claim that early adopters create them. The WEF Four Futures for Jobs in the New Economy sheds light on the uncertainty.

The report picks up on earlier work estimating 170 million new jobs will be created by 2030. Meanwhile, 92 million workers will be displaced. There is no reason to believe that those displaced fit seamlessly into the new positions.

So, we have a world of opportunity and a world of pain coexisting. Given the WEF’s evidence is skewed towards executives at large corporations, what can leaders at SMEs take from its analysis?

The conclusions will be clear to regular readers of The Profit Elevator. The report recommends starting with business bottlenecks rather than an abstract AI strategy. This complements other research linking success to hands-on executive involvement rather than the appointment of a head of AI.

Bottlenecks occur in repetitive, information-heavy processes that constrain growth. These include preparing quotations, customer support, scheduling, marketing production and bookkeeping preparation. Start with one or two where errors are easy to check and results are readily measurable.

Pilots should record metrics such as time saved, output produced, error rates and customer response times. They should also include time taken for checking or supervision and the software and running costs of AI.

The next step is to redesign workflows. For example, AI might retrieve information or draft content. A named person should then check its work and approve any consequential decisions. Always record errors and corrections and use them to improve the process.

Small businesses have an advantage here because they have fewer systems, shorter decision chains and are less likely to face employee resistance.

Businesses should also augment before considering automation. Aim to make capable employees faster by drafting for them, making recommendations or flagging anomalies for them to act on. It is better for AI to provide options than to commit the company to a single course of action.

When it comes to training, it is important to teach people to use AI in their jobs. Sales staff need to know how to validate AI-generated research on prospects. Marketers should know how to spot fabricated claims and sources. Finance staff must know what confidential data they can and cannot enter into an AI system. General AI-awareness courses will not teach these specifics.

It is also essential to protect the company’s proprietary data. CEOs should establish rules covering which tools can be used, what information should always remain outside of AI systems and who owns AI-assisted output. A small business needs an accountable owner, an approved-tools list and clear rules. It does not need a separate governance department.

The WEF report also advises firms to avoid being trapped in multi-year software contracts. Custom systems can become dependent on the consultants that designed them, while proprietary formats make data difficult to export. Short pilots, documented processes, exportable data and interchangeable models are the way forward.

The WEF’s Four Futures report is not a comprehensive guide to the future of AI. It presents four speculative scenarios without a clear steer on which is most likely. Readers will have their own views.

Business leaders must also consider regulation and court decisions, energy and computing costs, and potential customer resistance to the use of AI. As always, being close to your customers remains the most valuable skill at your disposal.

Most surveys about the extent of AI adoption focus on executive opinions. Typically, these are more bullish than the rank and file. They also tend to focus on profits rather than wages.

Twenty-four percent of executives surveyed in the WEF report expect AI to create jobs, while 45% believe it will improve margins. This suggests a greater emphasis on cost reduction than growth, which might alarm workers. But thinking about efficiencies is a considerable part of an executive’s job. The reality is that no one yet knows what impact AI will have on employment.

A rational policy for the CEO of an SME is to run two or three closely controlled pilots, put one senior person in charge and measure the financial and quality outcomes.

Then train staff on everyday workflows, keep consequential decisions for humans and take steps to protect proprietary data.

Over the longer term, review talent pipelines and development pathways with a view to replacing seniority-based roles with capability-based ones. Do not eliminate junior roles unless you have a clear alternative path to develop the next generation of leaders.

There is time to adapt to AI provided a business understands the issues and has a plan to deal with them. Reports such as those from the WEF provide useful background knowledge, but they cannot tell you where AI will create value in your firm.