“Apple buys smaller technology companies from time to time, and we generally do not discuss our purpose or plans.”
This boiler-plate public relations statement by Apple will be nothing new for those who have been watching the company’s artificial intelligence appetite steadily grow over the last five years.
It’s a statement they have had to issue with increasing frequency. According to Global Data, between 2016 and 2020, Apple bought more AI companies than any other in the game – including Google, Microsoft, Facebook, and Accenture.
While even non-tech companies the world over are in the midst of something of an AI gold rush, unsurprisingly it’s the big tech players who are out on the frontier. While Apple and its ilk made their trillions off the back of physical products like phones and laptops, these aren't resulting in as much revenue as they used to. If the future for these companies rests in the land of digital services, then putting AI development on the fast track is the logical next step in getting there.
The why of AI
According to Business Insider, 2019 was the first year since 2012 that the iPhone accounted for less than half of Apple’s revenue. Instead, services and digital products have begun to do more of the heavy lifting. Not only is the company vacuuming up AI startups, but it’s also attracting a lot of talent in the industry. This includes John Giannandrea, the former head of Google AI, who now resides at Apple as the Senior Vice President for Machine Learning and AI Strategy.
The New York Times reported that it was the right move for a company that, until that point, had lagged behind in the AI game. The machine learning-assisted voice recognition technology behind Google Assistant and Amazon’s Alexa is part of the reason why those two companies have beaten out Apple in smart speaker sales over the years.
Due to this fact and others, the company has begun to course-correct its AI strategy in a major way. Apple acquired the Dublin-based Voysis in early 2020, a startup whose work has focused on improving natural speech processing in virtual assistants. This is unsurprising, given that back in 2015, Apple acquired another speech-recognition company, called VocalIQ, according to the Wall Street Journal.
As we move into 2021, there are fewer and fewer features in phone, laptop, and tablet technology that are not infused with some kind of AI, and people like Giannandrea have been some of the driving forces behind this shift. Machine learning now powers a wide range of features in Apple devices. The company’s neural engine processor, for example, is designed to perform AI tasks such as working with image signal processors to improve the quality of digital photos in real-time.
The tech giants scrambling to buy AI startups are well aware of the fact that whoever does best at incorporating increasingly advanced AI into their products will be able to position themselves quite advantageously. With any AI application, the nature of machine learning and the fast pace of technological advancements means that being one step ahead of the competition is akin to being in an entirely different area code.
A bird’s-eye view of the corporate topography reveals that, even aside from the big tech titans, more and more companies are acquiring talented AI teams at ever faster rates. Even companies like Nike and McDonalds have bought AI startups in recent years to help them better measure consumer demand, and serve customers with voice-activated digital drive-thru agents.
An Apple a day
Not all AI applications are treated equally, however, and what’s telling is the data on the type of technology that’s being bought in these acquisitions, as it offers a window into how AI might significantly change certain industries in the coming years.
According to an analysis by CB Insights, the industries that the majority of these AI startups operated in prior to being acquired were most heavily concentrated in retail and healthcare, with applications focused on speech identification and cybersecurity.
These applications are most likely part of an effort to galvanize tech companies’ internal AI departments, but the ramping up of AI as it applies to health-related products and services is something to keep an eye on as the world slowly transitions into a post-pandemic era.
As CNBC reported just this month, Microsoft bought the speech-recognition firm Nuance for $16 billion, and unlike Apple’s opacity on these matters, it has made a point of broadcasting both the acquisition and its purpose. The company will be infusing its Microsoft Cloud for Healthcare with tools that Nuance has created over the years for use in recognizing and transcribing speech in voicemails, customer-service calls, and doctor visits.
The acquisition comes on the heels of the announcement that the company will no longer provide support for its voice-assistant Cortana on both iOS and Android. While it’s possible that Cortana will be getting an upgrade in the near future, thanks in part to Nuance’s contributions, Microsoft’s foray into the healthcare industry is a strong statement that it intends to keep up with Google, Apple, and others who are doing the same.
In a 2020 interview with Ars Technica, Giannandrea called attention to the fact that the company is making a big push to integrate machine learning into as many aspects of their products as possible, a number of which are health-related.
For example, the University Health Network in Toronto announced in a press release earlier this year that it would be launching a study that compares participant’s Apple Watch biometric data with standard physical testing to assess the efficacy of the technology. iPhones and Apple Watches lead in the smartphone and smartwatch market both in the US and worldwide, and that vast pool of available data may give the company an edge in breaking into the healthcare sector should users choose to share it.
Google may be Apple’s main competitor as it tries to do this, however. While the company has had a rough ethical go of it in recent years due to its involvement in developing AI for military applications, it’s betting that healthcare is on the cusp of a revolution at the hands of AI. As such, it has developed Google Health, a division aimed at using AI to improve disease diagnostics, treatment, and to better organize medical infrastructure by accelerating the digitization of medical records.
According to an announcement made by Google in November 2020, one logistical hurdle their AI systems are trying to overcome is the abundance of unstructured digital text (records that lack metadata to allow them to be easily mapped onto database fields). This kind of information requires a manual review of a particular document by a healthcare professional. That takes time and leaves room for human error.
Hence the company’s development of its Healthcare Natural Language API, which uses machine learning to identify “clinically relevant attributes based on the surrounding context [...].” The algorithm also “picks up the likelihood of a specific symptom or diagnosis, as captured in language nuances.”
This is a potential area of strength for Google. Just as Apple has been on an AI acquisition streak, Google has been, “doubling down on the number of research papers it publishes, opening more AI research centers around the world, and developing its own chips and hardware dedicated to running AI/ML processes,” according to a 2018 report by CB Insights.
The same report indicates that the health industry-focused subsidiaries of Alphabet, Google’s parent company, are honing in on ways to apply AI via its Verily Study Watch, a device similar to Apple Watch in that it collects biometric data.
None of this is to say that tech companies have had an easy time working their way into a heavily-regulated, slow-moving industry whose currency is patients’ sensitive personal information.
Microsoft, for one, readily acknowledges this difficulty. In a 2019 interview with The Financial Times, Eric Horvitz, the company’s chief scientific officer, said “We’ve continued, as a tech sector, to typically underappreciate a bottleneck [...] which comes from trying to move “from computer science principles to the real world of clinical care.” He goes on to say that tech companies are unfortunately rather good at ignoring the human elements that prevent the full adoption of proposed technologies.
Google has had a bumpy track record here as well. A few years after launching Google Health and Google Flu Trends back in 2008, for example, the company shuttered both operations. Regarding Google Health, they cited a vague lack of scalability in its operations as leading to its demise, but the company may have learned a valuable lesson while closing down its Flu Trends website.
In a 2015 blog post, Google’s AI team announced that it would still keep track of search data regarding the flu, but instead of providing that information directly to the public, it would work with institutions like Columbia University and the Centers for Disease Control and Prevention to help them develop models based on the data. This experience likely showed Google the potential that such partnerships could have on opening up the industry to them.
Fast-forward to August 2020, and Verily Life Services, Alphabet’s healthcare-focused subsidiary, is making a push into the insurance world, creating a subsidiary of its own called Coefficient. According to The Verge, “the company’s history in technology and health care could make the new Coefficient subsidiary an intriguing partner for insurers [...] its “analytics-based underwriting engine” will help employers better understand the risk that they’re taking on.”
Underpinned by a history of healthcare, the insurance industry may seem an unexpected target for Verily, but the move may be an example of just how versatile and unpredictable the world of AI applications can be.
The good, the bad, and the worrisome
If big tech companies can someday manage to use AI to streamline healthcare logistics, improve diagnostic capabilities, and even offer people better insurance options, the virtues of the technology are likely to be life-savingly obvious.
However, just as our society spent the last decade slowly waking up to the fact that its social media and consumer data was being harvested and sold to third parties, the risks associated with divulging medical information to the same companies who falsely offer assurances that their consumers’ data is safe are glaringly apparent.
It is possible that we will be confronted with a false choice in the near future, one between privacy and better health. Whether or not the companies behind the magic of AI choose to stick to the digital privacy principles they so often tout remains to be seen, and their behavior in how they use this powerful technology in the coming years will be the litmus test.