China’s Huawei has big ambitions to weaken the US grip on AI leadership

Ren Zhengfei, the reclusive founder and CEO of China’s embattled tech big, Huawei, is defiant about American efforts to impede his firm with lawsuits and restrictions.

“There is no such thing as a manner the US can crush us,” Ren mentioned in a uncommon current interview with international media. “The world can’t depart us as a result of we’re extra superior.”

It’d sound like bluff and bluster, however these phrases carry a measure of reality. Huawei’s expertise highway map, particularly within the area of synthetic intelligence, factors to an organization that’s progressing extra quickly—and on extra expertise fronts—than some other enterprise on this planet. Aside from its AI aspirations, Huawei is an ascendant player in the next-generation 5G wireless networking market, in addition to the world’s second-largest smartphone maker behind Samsung (and forward of Apple).

“The [Chinese] authorities and personal sector method is to construct corporations that compete throughout the total tech stack,” says Samm Sacks, who makes a speciality of cybersecurity and China at New America, a Washington assume tank. “That’s what Huawei is doing.”

Nevertheless it’s Huawei’s AI technique that can give it really unparalleled attain throughout the entire of the tech panorama. It would additionally elevate a number of recent safety points. The corporate’s technological ubiquity, and the truth that Chinese language corporations are in the end answerable to their authorities, are large the reason why the US views Huawei as an unprecedented nationwide safety risk.

In an unique interview with MIT Expertise Evaluation, Xu Wenwei, director of the Huawei board and the corporate’s chief technique and advertising and marketing officer, touted the scope of its AI plans. He additionally defended the corporate’s file on safety. And he promised that Huawei would search to interact with the remainder of the world to handle rising dangers and threats posed by AI.

Xu (who makes use of the Western title William Xu) mentioned that Huawei plans to extend its investments in AI and combine it all through the corporate to “construct a full-stack AI portfolio.” Since Huawei is a non-public agency, it’s difficult to quantify its expertise investments. However officers from the corporate mentioned final 12 months that it deliberate to greater than double annual R&D spending to between $15 billion and $20 billion. This might catapult the corporate to between fifth and second place in worldwide spending on R&D. In response to its web site, some 80,000 staff, or 45% of Huawei’s workforce, are concerned in R&D.

Huawei’s imaginative and prescient stretches from AI chips for knowledge facilities and cell units to deep-learning software program and cloud providers that supply a substitute for these from Amazon, Microsoft, or Google. The corporate is researching key technical challenges, together with making machine-learning fashions extra knowledge and vitality environment friendly and simpler to replace, Xu mentioned.

However Huawei is struggling to persuade the Western world that it may be trusted. The corporate faces accusations of intellectual-property theft, espionage, and fraud, and its deputy chairwoman and CFO (and Ren’s daughter), Meng Wanzhou, is presently below home arrest in Canada, awaiting attainable extradition to the US. America and a number of other different nations have banned the sale of Huawei’s units or are contemplating restrictions, citing issues that Huawei’s 5G tools that might probably be exploited by the Chinese language authorities to assault programs or slurp up delicate knowledge.

Xu defended the corporate’s status: “Huawei’s file on safety is clear.”

However AI provides one other dimension to such worries. Machine-learning providers are a brand new supply of threat, since they are often exploited by hackers, and the information used to coach such providers could include non-public data. Using AI algorithms additionally makes programs extra advanced and opaque, which implies safety auditing is more difficult.

As a part of an effort to reassure doubters, Xu promised that Huawei would launch a code of AI rules in April. This can quantity to a promise that the corporate will search to guard person knowledge and guarantee safety. Xu additionally mentioned Huawei needs to collaborate with its worldwide rivals, which would come with the likes of Google and Amazon, to make sure that the expertise is developed responsibly. It’s, nevertheless, unclear whether or not Huawei may enable its AI providers to be audited by a 3rd celebration, because it has finished with its {hardware}.

“Many corporations throughout the trade, together with Huawei, are growing AI rules,” Xu informed MIT Expertise Evaluation. “For now, we all know at the least three issues for sure: expertise needs to be safe and clear; person privateness and rights needs to be protected; and AI ought to facilitate the event of social equality and welfare.”

Stacking up

As Huawei advances in AI and progresses towards its intention of changing into a “full stack” firm, nevertheless, it might more and more appear too highly effective for a lot of within the West.

Already, it boasts a dizzying array of choices. Final 12 months, Huawei launched an AI chip for its smartphones, referred to as Ascend, that’s akin to a chip discovered within the newest iPhones, and tailored for working machine-learning code that powers duties like face and voice recognition. The expertise for the chip got here from a startup referred to as Cambricon, which was spun out of the Chinese language Academy of Sciences, however Huawei not too long ago mentioned it might design future generations in-house.

Huawei additionally sells a spread of AI-optimized chips for desktops, servers, and knowledge facilities. The chips lag behind these provided by Nvidia and Qualcomm (each US corporations) when it comes to sophistication, however no different enterprise can boast such a spread of AI {hardware}.

Then there’s the software program. Huawei gives a cloud computing platform with 45 completely different AI providers—related in scope to choices by Western giants like Google, Amazon, and Microsoft. Within the second quarter of 2019, Huawei will even launch its first deep-learning framework, referred to as MindSpore, which can compete with the likes of  Google’s Tensorflow or Fb’s PyTorch.

AI can be woven into Huawei’s ambitions to supply the 5G tools that can join every little thing from industrial equipment to self-driving vehicles. “We have to use AI to cut back upkeep prices,” Xu mentioned. “Telecom networks have gotten increasingly advanced—70% of community failures are brought on by human errors, and if we use AI in community upkeep, over 50% of potential failures will be predicted.” 


Xu’s statements on AI ethics are additionally, in a way, a part of an effort to steer the world’s AI improvement. Guaranteeing moral AI will imply crafting technical requirements, which shall be vital to shaping the way forward for the expertise itself. America has exerted an outsize affect over its improvement of the web by technical requirements. 

To that finish, the Chinese language Affiliation for Synthetic Intelligence, a state-run group, arrange a committee earlier this 12 months to draft a nationwide code of AI ethics. A number of of China’s large tech corporations, together with Baidu, Alibaba, and Tencent, even have initiatives devoted to understanding the influence of AI.

Agreeing on AI ethics and requirements might show a problem as tensions between East and West escalate, nevertheless. Various nationwide governments, in addition to organizations just like the EU, are additionally searching for to set the principles of the highway. “AI brings worth in addition to issues and confusions,” Xu informed MIT Expertise Evaluation. “World collaboration is required to handle these issues.”

And worldwide collaboration shouldn’t be precisely a forte of the US proper now. Certainly, outdoors of its personal borders, the American authorities can do solely a lot to hamper Huawei. Some allies are apparently tiring of US strong-arm ways; the UK and Germany each appear more and more unlikely to ban Huawei from supplying 5G tools and different services and products. 

The corporate’s curiosity in ingratiating itself with cautious nations additionally has its limits. In current feedback its CEO, Ren, contended that the worldwide image is altering, at the least in technological phrases. “If the lights exit within the West, the East will nonetheless shine,” he mentioned. “And if the North goes darkish, there may be nonetheless the South. America doesn’t signify the world. America solely represents a portion of the world.”

Both manner, there shall be Huawei.

Source link


Please enter your comment!
Please enter your name here