Recently, McKinsey released a global survey report on the status of AI, which is the third year in a row. Interviews with executives and a survey of practitioners show that the gap between companies that use AI and those that don’t may widen.
The survey report shows that in the field of technology and telecommunications, the use of artificial intelligence is more common than in other industries, followed by the automobile and manufacturing industries. More than two-thirds of respondents said adopting AI increased their income, but less than a quarter saw a significant impact.
McKinsey’s AI status report, together with questions about AI application and implementation, studies companies whose AI application leads to an increase of 20% or more in EBIT (interest and tax profit) in 2019. Compared with other companies, these companies may be more efficient in rating senior managers and are more likely to employ data scientists.
Compared with other companies with a difference of 20% to 30% or more, high-performance companies are also more likely to have strategic vision and AI roadmap, use AI model deployment or need to use comprehensive data in case of insufficient data. These results appear to be consistent with Microsoft’s altimeter group survey in early 2019, which found that half of high growth companies plan to implement AI next year.
If there is anything surprising in the report, it is that only 16% of the respondents said that their company has developed deep learning projects beyond the experimental stage. (this is McKinsey’s first year of focusing on deep learning.)
It is also surprising that the report shows that enterprises have made little progress in coping with the risks associated with AI deployment. Compared with the responses submitted last year, companies taking measures to mitigate such risks have increased by an average of 3% in responding to 10 different risks, from national security and personal safety to compliance and equity.
Cyber security is the only risk that most respondents say their companies are trying to address. There are a number of categories that believe that AI risk related to companies is declining, including equity and equity, from 26% in 2019 to 24% in 2020.
“Although some risks such as personal safety apply only to specific industries, it is difficult to understand why a high proportion of respondents do not recognize the general risks.” Roger Burkhardt, a partner at McKinsey, said in the report that “given concerns about racial prejudice and other examples of discriminatory treatment, such as age-based positioning in job advertisements on social media, it is particularly surprising to see that this risk has hardly been mitigated or improved.”
Not surprisingly, the survey found improvements in automation in some industries during the pandemic. VentureBeat found that this was true in industries such as agriculture, construction, meat packaging and shipping.
“The majority of high performing respondents said that their organizations have increased their investment in AI in each major business function in response to the epidemic, while less than 30% of other respondents said the same,” the report wrote
McKinsey’s global survey on AI in 2020 was conducted online from June 9 to June 19, and nearly 2400 results were obtained. 48% of the respondents said that their companies used some form of AI. A McKinsey survey of about the same number of business leaders in 2019 found that while nearly two-thirds of companies reported increased revenue due to the use of AI, many are still trying to expand their use.
Another state of AI
A month before McKinsey released its business survey report, airstreet capital released its “state of artificial intelligence” report, which has continued into its third year. The London based venture capital firm found the AI industry popular in corporate finance, but reported that the concentration and calculation of AI talent was “a huge problem.”.
Other serious problems identified by airstreet capital include the continued brain drain from academia to industry, and the replicability of models created by small companies. A team of 40 Google researchers recently found that inadequate specifications are a major obstacle to machine learning.
Many of the conclusions in the report and the latest analysis of AI research papers found that deep learning activities among large technology companies, industry leaders and elite universities are becoming more and more concentrated, which exacerbates inequality. The team behind the analysis said that the growing “computing gap” could be solved to a certain extent through the implementation of national research cloud.
As the end of the year approaches, we can expect more reports on the state of machine learning. AI status reports released in the past two months show challenges, but say AI can help businesses save costs, generate revenue and follow proven best practices to succeed. At the same time, researchers are looking for solutions to the huge opportunities associated with deploying AI.
Editor in charge: CC