OECD: AI threat to trust in business…
The promise is there, but AI’s limitations are also coming increasingly to the fore, potentially undermining trust and leaving some questioning the real value of the technology.
OECD Business and Finance Outlook 2021: AI in Business and Finance says AI has the potential to provide insights, recommendations and efficiencies for sectors ranging from health care to agriculture, but it also can create new risks or reinforce existing risks, including eroding competition, privacy and the right to fair, equal treatment.
It’s already ‘transforming’ financial services, with financial markets spending up large in 2020 with global spend of more than US$50 billion in AI. But the OECD report says risks include entrenching bias, lack of explainability of financial decisions, introducing new forms of cyber attacks and automating jobs ahead of society adjusting to the changes.
“Fewer than two out of five of companies that had made an investment in AI reported business gains.”
Also flagged are challenges related to privacy, autonomy, transparency and accountability.
Complex AI algorithms which are difficult or even impossible to explain are also cause for concern, presenting challenges for trust and accountability.
“AI applications offer remarkable opportunities for businesses, investors, consumers and regulators,” the report says.
“AI can facilitate transactions, enhance market efficiency, reinforce financial stability, promote greater financial inclusion and improve customer experience.”
But on the flip side, the report notes inappropriate use of data or use of poor quality data can create or perpetuate biases and lead to discriminatory and unfair results and herding behaviour.
“Growing dependencies on third party providers and outsourcing of AI models or datasets raise issues around governance and accountability, while concentration issues and dependence of a few large dominant players may also arise, given the important investment required for the deployment of AI techniques is based on in-house capabilities, rather than outsourcing.
“Existing model governance frameworks may be insufficiently addressing risks associated with AI, while the absence of clear accountability frameworks may give rise to market integrity and compliance risks.”
Those concerns have seen the European Commission recently publish its proposed Artificial Intelligence Regulation – the first legal framework intended to govern the use and development of AI.
Stanford University’s 4th AI Global Index report showed businesses invested $67.9 billion in AI last year, a 40 percent year on year increase.
But despite that investment, numerous reports suggest companies are having trouble implementing AI, let alone getting any real value from AI solutions. An MITSloan report noted that fewer than two out of five of companies that had made an investment in AI reported business gains. Even among those making ‘significant’ investment, business gains were only seen by around 60 percent.
A McKinsey report too, highlights a lack of commercial use of AI, noting its latest findings show no increase in AI adoption, though some companies ‘are capturing value from AI at the enterprise level, and many are generating revenue and cost reductions at least at the function level’.
Some companies, however are seeing value accrue, with 22 percent of those surveyed by McKinsey saying that more than five percent of their enterprise-wide EBIT in 2019 was attributable to their use of AI.
“It’s clear that we’re still in the early days of AI use in business, with less than a quarter of respondents seeing significant bottom-line impact,” says Michael Chui, McKinsey Global Institute partner. “This isn’t surprising – achieving impact at scale is still elusive for many companies, not only because of the technical challenges but also because of the organisational changes required.
“However, those seeing AI contribute more than 20 percent to earnings before interest and taxes are not just from the tech sector. So it is possible for any company to get a good amount of value from AI if it’s applied effectively in a repeatable way,” he says.
“We do think AI is worth the investment, but it requires effective execution to generate significant value, particularly at enterprise scale,” Chui says.
Irving Wladawsky-Berger, a research affiliate at MIT, highlighted three key challenges: The need for large amounts of data to train algorithms accurately and associated issues of bias; the need for considerable computing power; and that deep learning ‘is quite shallow’ – capable of powerful pattern recognition, but struggling with reasoning, generalising from rules they discover and ‘common sense’, creating effectively an ‘idiot savant’ which excels at very clearly defined set tasks, but can get things very wrong if faced with unexpected inputs.
“Like other transformative technologies, AI is both powerful and limited,” Wladawsky-Berger says.
