Over the past twenty years, China has emerged a top contender for global AI supremacy. At the forefront of this endeavor is Alibaba.

Though Alibaba was originally founded as an e-commerce site back in 1999, since then it has evolved into a multinational technology company that provides consumer-to-consumer (C2C), business-to-consumer (B2C), and business-to-business (B2B) sales services via web portals, as well as electronic payment services, shopping search engines and cloud computing services.

However, AI is where the company has set its sights on next.

In addition to offering a number of AI solutions such as platforms and cloud services, Alibaba has launched dozens of highly interesting and innovative AI projects. Here is a look at some of the ways the company is using AI and ML to boost competitive advantage. 

AI to Optimize e-Commerce Supply & Demand

Alibaba’s massive volume of first-party sales also provides them with a significant strategic advantage: more data than any other retailer in the world including Amazon and Walmart. Since the dawn of the millennium, they’ve been able to use this data to build some of the fastest and most accurate AI algorithms in existence. 

One of Alibaba’s most high profile AI success stories was the 2020 Singles Day shopping bonanza. For those who may be unfamiliar, back in 2009, Alibaba invented a new, unofficial holiday now known as Singles Day where singles treat themselves to a gift on November 11. 

Over the years, Singles day have blown up from a small, niche event into the largest physical retail and online shopping day in the world. In fact, from November 1-11, 2020, Alibaba generated $74.1 billion in sales, 26% more than the previous year. 

In order to accommodate the tsunami of orders expected to come in, Alibaba needed to reimagine its AI algorithm. Plus, like many organizations, after months of irregular shopping habits, Alibaba’s existing algorithms were starting to malfunction. A complete rethink was in-store.

According to MIT Technology Review, in response, “Alibaba doubled down on its short-term forecasting strategy. Rather than project shopping patterns based on season, for example, Cainiao [Alibaba’s logistics arm] refined its models to factor in more immediate variables like the previous week of sales leading up to major promotional events, or external data like the number of covid cases in each province.” 

As live streaming had taken off during the pandemic, they also built a new forecasting model to project what happens when popular live-stream influencers market different products.

Alibaba has also paved the way when it comes to natural language processing (NLP). Alibaba’s customer service chatbot, AliMe, uses speech recognition, semantic understanding, personalized recommendations, and deep learning to direct people to the best-fit product or service. According to HBR, during the Single’s Day shopping festival in 2019, AliMe responded to 300 million queries accounting for 97% of the customer services on its e-commerce platforms. To put this in perspective, it would have taken 85,000 human service staff to complete the same amount of work. 

Last but not least, another way they capitalized on the company’s vast trove of e-commerce data to build a new fraud identification system capable of identifying counterfeit items with 96% accuracy. 

According to the Global Times, “Chinese e-commerce giant Alibaba recently updated its AI platform and made it able to identify authentic and counterfeit logos of luxury brands in 30-50 milliseconds. The speed is 10 times quicker than the speed people blink, industry media reported Wednesday. The platform was built by the Alibaba Turing lab and it contains a database of more than 1 million logos covering 500 luxury products categories, which is the biggest database of this kind in the world, according to the report. The AI anti-counterfeiting platform contains data from about 13.7 billion picture samples.”

The M6 Mega-Transformer

In November of 2021, Alibaba’s science and technology research institute, Damo Academy announced that its “Multi-Modality to Multi-Modality Multitask Mega-transformer” (M6) artificial intelligence (AI) system increased its number of parameters from 1 trillion to 10 trillion. This far exceeds the trillion-level models previously released by Google and Microsoft making the M6 the world’s largest AI pre-training model.

In addition, M6 has achieved “ultimate low carbon, high efficiency in AI models” by only using 512 graphic processing units (GPU) to train a 10 trillion parameter neural network within ten days. In fact, M6 achieved the same parameter scale as OpenAI’s GPT-3 with only 1% of its energy consumption.

According to reporting by Verdict, “The new thing about M6 is that it has hundreds or thousands of times the number of “neurons” compared to other AI systems currently being trialed, perhaps enabling a learning ability that is more like the human brain. According to Alibaba, M6 has been applied in over 40 scenarios, with a daily parameter volume in the hundreds of millions.”

As promising as all this sounds, it should be noted that some AI experts have expressed skepticism regarding M6’s impressive performance stats in the past

Become an AI Data Analytics NETWORK member today!

Premium content. Expert insight. Instant access. All for FREE.