AI technology is becoming increasingly powerful, and, with much of the funding coming from the Tech States, several scientists have raised the alarm about the role corporate interests play in its development. How worried should we be?
Here’s what you need to know this week:
- AI researchers questioned corporate influence over their work
State by state:
- Apple extended its lead in wearable technology
- Google pretended to be the planet Pluto
- Facebook downgraded misinformation spreaders
- Tencent plotted to grab a bigger stake in Chinese gaming
- Amazon moved further into healthcare
- Microsoft faced scrutiny over its treatment of female employees
As AI technology becomes more powerful, how far should big tech be involved in driving forward innovation? Companies like Google and Microsoft provide much-needed support for AI researchers and start-ups, who typically need both cash and cloud computing grunt to succeed.
But some are worried that the increasing privatisation of AI could constrain its societal benefits. Are commercial motivations consistent with research independence and with the development of AI for the public good?
Recent developments have thrown a spotlight on the issue.
- AI researchers split from OpenAI. OpenAI was founded in 2015 with the aim of developing general-purpose AI safely and for the “benefit of humanity”. In 2019 it accepted $1 billion from Microsoft, in exchange for the right to commercialise its research. The investment caused internal tensions, provoking several researchers to break away, the Financial Times reported. They’ve formed a new AI startup called Anthropic – which aims to create “safe AI” – and have already raised $124 million from figures like Jaan Tallinn, the Estonian computer scientist who co-founded Skype.
- DeepMind insisted its partnership with its parent company, Alphabet, has been “extraordinarily successful,” despite reporting from the WSJ that the AI research group had failed to win more autonomy from Google over its work. A DeepMind spokeswoman told us that the company already enjoyed “operational autonomy” and had Alphabet’s full support. Demis Hassabis, DeepMind’s co-founder, will be speaking at Tortoise’s Responsible AI Forum at Waddesdon on 10 June.
- Academics found that AI Research was increasingly privatised. A recent study found that Google-funded AI research accounted for 167 of the papers accepted at a recent major conference – more than twice the number from the second most represented institution (Stanford). Big names in the field, like Geoffrey Hinton and Ian Goodfellow, have switched from academia to industry. “Once scientists from corporations begin to dominate the research enterprise,” the researchers write, “it will simply be too late.”
Is there a way to square the circle? The breakaway OpenAI researchers have registered their new company as a public benefit corporation, with special governance arrangements to help it “maintain advanced AI for the benefit of humanity”.
They will form a “benefit committee” made up of people with no connection to the company who will have the final say on critical matters. Such oversight arrangements at AI companies may turn out to be more important than higher-profile equivalents at companies like Facebook.
Following Bill Gates’ “inappropriate” affair with a colleague, the tech state is facing broader scrutiny over how it treats women. Maria Klawe, a former Microsoft board member, claims she was “silenced” after publicly questioning its CEO Satya Nadella in 2014. Nadella had just told 8,000 conference attendees that women should rely on “good karma” to get pay rises and Klawe, who was moderating the session, received a cheer when she argued that women should negotiate instead. Klawe now claims that Microsoft chairman John Thompson asked her to step down from the board following the conference because she had apparently made the company look bad.
Facebook’s Oversight Board is the analogue way the tech state deals with moderation issues. But the company is also making important UI (user interface) changes which could have a more significant effect. Last week it announced that users who regularly post misinformation will be made less visible on the newsfeed. Facebook already demotes false posts, but this is the first time its algorithms will discriminate against people who spread misinformation. Watch out Robert F Kennedy, who in February we identified as one of the biggest superspreaders of misinformation on the site.
Amazon may have inked its second biggest acquisition since Whole Foods when it acquired James Bond studio MGM for $8.45 billion last month, but healthcare is where the tech state is looking to make big strides. Amazon is developing at-home medical tests including a Covid-19 home testing kit as part of a new line of products to be branded “Amazon Diagnostics”, Business Insider reported. Not only that – it’s considering opening bricks-and-mortar retail pharmacies in the US, either through stand-alone stores or using its Whole Foods network. Amazon has also secured a “Amazon Pharmacy” trademark in the UK.
The future of search is conversation – according to Google. Last month it demonstrated an impressive new voice search technology which the tech state said “showed what it’s like to have a chat with Pluto”. Speech-driven search is already in products such as Google Assistant and Echo. But there’s a big difference between answering a simple command like “play music” and a complex query of the sort we type into Google Search. As pointed out in The Verge, however, closing this gap is tough: despite major improvements in recent years the technology still struggles with the complexity of the real world.
Last week Apple quietly unveiled a new piece of technology: AssistiveTouch allows an Apple Watch wearer to control her device without touching it (you can stop a timer, for instance, by clenching your hand twice). As tech blogger Neil Cybart points out, such R&D driven advancements show how far Apple has extended its lead in wearables. In 2021 it’s on track to sell more than 100 million wearable devices and is now seeing more people buy a wearable than buy a new iPhone for the first time.
Do Chinese antitrust regulators really want to crack down on Tencent, or do they just want to flex their muscles? Since December, authorities have delayed a merger of China’s two biggest video game live-streaming platforms, Huya and DouYu, over concerns that it would give Tencent too much power. The tech state already owns 43 per cent of China’s video gaming market and would own 67 per cent of the merged enterprise, the South China Morning Post reported. But signs are that China will allow the deal to take place. Chen Shaojie, DouYu founder, told investors on 18 May that the merger “is on track”.
And finally…The Responsible AI Forum at Waddesdon
On 10 June Tortoise and Lord Rothschild are hosting an absolutely stellar event with the aim of setting out an agenda for the development and deployment of AI as a responsible technology. Keynote speakers include Masayoshi Son, founder of Softbank, Demis Hassabis, founder of Deepmind, and Helle Thorning Schmidt, former prime minister of Denmark and co-chair of the Facebook Oversight Board – among many others.
Thank you for reading the Tech States Sensemaker. We’re trying to do something new with our tech coverage – and we’d love to hear what you think. Please email opinions, tips or stories to email@example.com or firstname.lastname@example.org.