Decades ago, all watches and clocks had hands. These timepieces only became analog when their digital successors gained prominence.
With the advent of artificial intelligence (AI), we might think of its inspiration and predecessor, human thought, as “organic intelligence” (OI). Further, we can look at AI as something with the power and flexibility to extend and enhance OI in many fundamental ways, especially when supported by comprehensive, inclusive training.
AI: A General-Purpose Technology (GPT) In the Making
In his keynote address at the Huawei Connect 2018, Rotating Chairman Eric Xu said Huawei recognizes AI as “a combination of technologies that, together, form a new general purpose technology (GPT).” He goes on to cite the award-winning 2006 book Economic Transformations: General Purpose Technologies and Long-Term Economic Growth. In that work, authors Richard G. Lipsey, Kenneth I. Carlaw and Clifford Bekar declare GPTs “have transformed the economic, social, and political landscape in the 10,000 years since the Neolithic Agricultural Revolution.” They also delineate the following four criteria that define a GPT.
- It is a single, recognizable generic technology.
- It initially has much scope for improvement but comes to be widely used across the economy.
- It has many different uses.
- It creates many spillover effects.
The authors identify 24 processes, products, and organizational systems as GPTs, ranging from domestication of plants and animals in the Neolithic era to business virtualization in the 21st century. More recently, the authors have added two other GPTs to their list: nanotechnology and AI.
Fulfilling the Promise of AI: Technologies and Training
The rapid growth and broad range of commercially available AI-powered solutions for customer support and information access demonstrate AI’s potential to become a GPT. But there are important questions to be answered first.
One such question is how to ensure the training of AI tools and solutions accurately and comprehensively reflects the full diversity of all relevant OI. Since training requirements will vary by audience and application, to deliver maximum value, AI solutions must also be easy to tailor for specific applications and audiences. However, those solutions and their trainers must also avoid preemptively or unknowingly excluding or under-representing relevant information or constituencies.
These requirements demand two things from developers of AI technologies: powerful computing platforms and comprehensive, flexible, and inclusive training solutions. Fortunately, the continuing evolution of multiple underlying technologies are enabling AI tool builders to deliver power and flexibility to creators of AI-powered solutions.
In that 2018 keynote, Chairman Xu identified multiple gaps between the “stellar achievements” and “lukewarm adoption” of AI. “To close these gaps, we need the right technology, the right talent, and the right industry ecosystem,” Chairman Xu said. He added that readily available, affordable computing power and better, faster training of AI models are critical to achieving these goals.
Huawei is committed to delivering AI tools and OI support that address these needs effectively.
At Huawei Connect 2019, the company introduced the Atlas 900, the world’s fastest AI training cluster. This offering harnesses the power of thousands of Huawei’s Ascend AI processors and powers a range of cloud-based, AI-powered services. The power and flexibility of the Atlas 900 will help make AI training faster, easier, more accurate, and more inclusive.
At the same event, Huawei also announced 43 new AI cloud services powered by Ascend processors. These include Elastic Cloud Servers (ECSs), autonomous driving services, a knowledge graph creation and management service, and significant performance upgrades to Image Search and Content Moderation services.
In November 2019, Huawei and China’s Peng Cheng Lab jointly released Peng Cheng Cloud Brain II Phase 1. Built atop the Atlas 900 AI Cluster, Peng Cheng Cloud Brain is intended to offer performance as high as 1,000 petaFLOPS. A single petaFLOP is the equivalent of a thousand trillion floating point operations per second. Access to such performance levels can extend and support the OI of AI developers and researchers, plus enable more and better inclusive training solutions.
The GSM Association (GSM) awarded the Atlas 900 its Global Mobile (GLOMO) Tech of the Future Award in February 2020. Judges cited the platform’s “incredible speed and performance” and “minimal carbon footprint.”
That same month, the HUAWEI CLOUD business unit, Huazhong University of Science & Technology, and Lanwon Technology joined the fight against the COVID-19 pandemic. They developed and launched a COVID-19 medical image analysis service. The service uses Huawei AI (artificial intelligence) computer vision and image analysis technologies and Ascend AI chips to deliver quantification results quickly and accurately to imaging and clinical doctors. They can then accurately diagnose COVID-19 and make informed quarantine decisions faster. Huawei is offering the service to hospitals in need at no charge.
Huawei is also investing in the OI that drives AI development. In January 2020, Huawei launched its inaugural Artificial Intelligence of Things (AIoT) Innovation Bootcamp in Singapore. This event brings aspiring students together to learn, experiment with, and explore AIoT technologies and applications. To further its commitment, Huawei announced plans to invest an additional US$1.5 billion in its developer program over the next five years. The goal is to grow the program to support five million developers among Huawei’s worldwide partners.
These investments and activities, along with those of other companies and institutions around the world, will accelerate AI development and make it more of a true GPT. Such efforts will also ease and speed the development of inclusive, comprehensive training that AI tools and solutions need to maximize their accuracy and value. In time, “inclusive training” may come to be seen as “the new IT.”
Note: A version of the above content originally appeared at the Huawei blog site. Also, on April 29, Huawei will host a webinar entitled “COVID-19, Connectivity and Community.” Speakers will include Huawei USA SVP Joy Tan and VP Scott Jamar and Debra Ruh, CEO of Ruh Global IMPACT. They will highlight what Huawei and other organizations around the world are doing to connect and support everyone affected by the COVID-19 pandemic. Click here or on the image above for more information and to register for this online event. And to learn more about the “latest and greatest” Huawei technologies, check out HDC.Cloud, the Huawei Developers Conference, online at https://www.huaweicloud.com/intl/en-us/HDC.Cloud.html.