Artificial intelligence is expected to contribute a potential US$15.7 trillion to the global economy by 2030, so it’s not surprising that more companies are considering buying an AI firm to bolster their tech capabilities. The number of AI deals rose from only 10 in 2012 to 166 in 2018, according to CB Insights, and there were more than 140 AI acquisitions in just the first eight months of 2019.1
If your company is thinking about jumping on the AI bandwagon, you should know that this deal won’t be like others you’ve made in the past. So you need to prepare.
The reality of the market
For one thing, the maturity of these companies varies markedly. Six out of 10 of the approximately 1,600 European AI startups in MMC Ventures’ 2019 State of AI study are only at the earliest stages of AI implementation, while only one out of six has reached the growth stage.2 Buying an AI company that’s in the early stages might not provide the mature solution you seek. In fact, 40 percent of the surveyed companies that claimed to be AI startups actually had very little AI tech — a concept called ‘AI washing,’ or ‘fake AI’.3
In addition, nine out of 10 AI startups in the study were found to address only a specific business function or sector, while just one out of 10 offers a sector-agnostic capability. This ratio has remained stable, and the proportion of sector-agnostic AI firms is expected to remain modest, so companies that want to buy firms with broad AI capabilities need to investigate the range of products offered. Before signing a deal, companies should understand the AI product’s scope and ensure that it’s a good fit for its business model, customer base, and specific needs.
Furthermore, AI algorithms don’t work effectively without the right kind of data, and some startups may over-promise the scope, accessibility, and uniqueness of their data assets. It’s vital to evaluate not just the startup’s AI technology, but also the quality and quantity of the data it owns. To judge that effectively, decision-makers should ask product managers, data scientists, and technologists to assess how the startup’s data compares with its competitors’.
Being responsible is critical
Another unique AI aspect to consider when evaluating a deal concerns responsible AI. Any business using this technology may be asked to explain how an AI model or algorithm reached a particular decision — and whether that decision was fair and ethical.
Has the company you’re thinking of acquiring developed responsible AI principles and practices? It’s important to ask the company’s management, data scientists, and others with knowledge of the AI system to address this question and provide explanations that are understandable to different stakeholders: consumers, regulators, and business sponsors. Taking this step will build employee and customer confidence in the AI technology you are using, helping ensure the value of your acquisition.
After signing on the dotted line
Of course, the work doesn’t end when a deal closes. AI, by its nature, creates inherent challenges around trust and accountability with customers, governments, employees, and other stakeholders. The plan you develop should do more than simply outline when and how your and the AI firm’s assets and operations would be consolidated. It should also explain how the revised operation would stand up to AI regulations and social standards.
To deal with these ethical challenges, you should develop an integration plan early on — one that includes ways to apply responsible AI practices. The plan should address the principles of responsible AI: ethics and regulations, robustness and security, interpretability and explainability, and bias and fairness.
As two companies combine key data that’s used to power the AI system, it is important to coordinate compliance departments in order to address both ongoing and developing rules around AI, including data and privacy laws. A robust governance system should also be put to places to determine how the company will use AI.
The AI bottom line
All of these challenges — limited AI capabilities, poor-quality data, and a lack of AI principles and practices — should prompt you to evaluate AI companies through a different lens. But a growing number of firms are deciding that AI’s opportunities outweigh its challenges. Is your business one of them?
