The surveys, studies, forecasts and other quantitative assessments of the health and progress of AI for this week find increased VC excitement over AI startups, low trust in AI advice by U.S. consumers and global business executives, China and Germany ahead of the US in the use of AI in healthcare while the US spends more, and the inaccuracy of facial recognition.
Funding to AI startups worldwide in Q2 of 2019 totaled $7.4 billion in 488 deals, up from $5.2 billion in Q1 [CB Insights]
$803 million were invested in the first half of 2019 in 43 VC deals worldwide involving the application of AI and machine learning to cybersecurity; last year’s total was $1.7 billion [KPMG]
Only 46% of senior executives say that AI is one of the top 3 important topics for their companies over the next 3 years; cybersecurity is the top rated topic at 73% [Deloitte survey of 1,048 senior executives]
Only 39% of senior executives say their company has a strong cultural orientation to data-driven insights and decision-making, and only 37% feel employees in their company are aware of the importance of data analytics [Deloitte survey of 1,048 senior executives]
49% of U.S. consumers would trust AI-generated advice for retail, 38% would trust AI-generated advice for hospitality, such as checking/comparing flight or hotel options or restaurant recommendations, while only 20% would trust AI-generated advice for healthcare and 19% for financial services [Invoca/The Harris Poll survey of 2,048 U.S. adults]
33% of US healthcare professionals have implemented AI into their practice, compared to a 15-country average of 46%; countries like Germany (41%) and China (85%) surpass the U.S. in the use of AI technologies in healthcare, despite the fact that the U.S. has one of the highest spends of AI in healthcare for preliminary diagnosis per capita at $0.06, while China’s spend is $0.002 per capita and Germany’s is $0.03 per capita [Philips Future Health Index 2019 survey of more than 3,100 healthcare professionals in 15 countries]
37% of US healthcare professionals associate AI in healthcare with less human interaction and only 20% associate AI with more accurate diagnosis [Philips Future Health Index 2019 survey of more than 3,100 healthcare professionals in 15 countries]
Researchers at Memorial Sloan Kettering Cancer Center used a deep learning model to analyze a dataset of 44,732 whole slide images from 15,187 patients, detecting prostate, skin and breast cancer with close to 100% accuracy, allowing pathologists to focus on the 25%-35% most critical slides [Nature Medicine]
Only 4% of hospital and health system boards have members with direct technology experience relevant to healthcare technology; 76% of board members said their hospital could not identify new technologies, including artificial intelligence, machine learning or patient privacy monitoring, to understand the risk-benefit equation of their spending choices [Black Book survey of 494 C-Suite hospital executives]
A data set of 7,500 real-world, unmodified and naturally occurring images managed to reduce the accuracy of AI by 90%, exploiting “deep flaws in current classifiers including their over-reliance on color, texture, and background cues” [arxiv.org]
59% of organizations worldwide have already deployed AI and they expect to double the number of AI projects in place within the next year; top reasons for deploying AI are customer experience (40%) and automating tasks (20%); top challenges are a lack of relevant skills (56%), understanding AI use cases (42%), and concerns with data scope or quality (34%) [Gartner survey of 106 Gartner Research Circle Members]
53% of financial services executives see new technologies such as blockchain, AI, machine learning and IoT as the number one trend impacting their companies; 83% believe that new entrants could become significant competitors—just 3 years ago only 7% viewed fintech as a threat [NTT Data survey of 471 Financial Services executives in the U.S., U.K., Germany, Spain, Italy and Japan]
60% of senior executives believe that decisions made by AI are fairer than those made by people because people are subject to bias; 52% said they could not trust AI because the data used may be incorrect or biased; 60% could not trust decisions made by AI on its own and preferred that the final decision be made by a person; 16% would trust AI rather than a human being to recommend a financial investment [Fujitsu survey of 900 CxOs and decision-makers in the US, the UK, Spain, France, Germany, China, Singapore, Australia, and Japan]
65% of IT executives say that machine learning is now “extremely important” or “very important” for network management, up from 45% in 2018 [Kentik survey of 388 attendees at the Cisco Live US conference]
The AI in construction market worldwide is forecast to reach $4.51 billion by 2026, up from $429.2 million in 2018 [Reports and Data]
The AI in agriculture market worldwide is forecast to reach $980 million by 2024 [Intelligence Journal]
By 2023, 80% of organizations using AI for digital commerce will achieve at least a 25% improvement in customer satisfaction, revenue or cost reduction [Gartner]
130,000 people across 60 countries in 50 different languages have worked on crowdsourcing (training AI) tasks for DefinedCrowd since it launched in 2015 [Seattle Times]
You can be correctly identified in an “anonymized” database in the U.S. 81% of the time (on average) with just your zip code, gender, and date of birth [Nature Communications]
Researchers from the University of Essex found that the Metropolitan Police’s facial recognition technology is 81% inaccurate—in the vast majority of cases it flagged up faces to police when they were not on a wanted list [Sky News]
“Well, I think if you can get some sort of readout that is, you know, available from the unit in your apartment, the status of where you are today, to be interactive in a sense, broadcasting the information that is … is collected about you and be analyzed by the artificial intelligence obviously to give you some kind of status, you know, you … you’re doing okay today or … or you ate too much yesterday” [One of the participants in focus groups with older adults administered by researchers at UC San Diego]
HT: Becker’s Hospital Review, ImportAI Newsletter, Benedict Evans Newsletter
