The fourth industrial revolution is driving society into an intelligent world. Artificial Intelligence (AI) is one of the key tech pillars amid this digital transformation. A few years ago, a fancy picture was projected for AI, where many experts saw its development exceeded expectation.
Undoubtedly, AI has been changing our world since its arrival, as it has quickly woven into almost everything. More often than not, it performs much better than expected.
The excitement, however, has shown signs of slowing down since the boom, where other technologies such as 5G, Wi-Fi 6 and storage related solutions have been witnessing strong growth in both demand and advancement.
There are many factors holding off the pace of such amazing technology to develop, but they mainly fall into two categories: the technological advancement, and the realisation of such demand in real life or the actual market landscape. The latter is usually related to cost as well as the doubt on its security and safety.
Technological tailback drags AI development
The advancement of AI is partly driven by the deep learning mechanisms. However, barriers are emerging around the mechanisms, mainly due to the lack of the computing power required for the improvements. The shortfall comes in both on the “soft” side and the “hard” side.
On the software front, there is a pressing need to break the theoretical blocks by a concrete boost in algorithm development and careful planning on resource allocation for computing power. For hardware, the availability and affordability of computing power has created the bottleneck.
Therefore, we must have a good set of solutions to address to challenges involving architecture, the actual production of chips and the implementation of software on chips and systems.
However, even resolving the technological issues doesn’t guarantee a sustainable growth for AI in real life. How to recognise the actual demand and the realistic market landscape has become a more important issue, especially when monetary terms are involved.
Investment return for AI isn’t reflected directly
One of the biggest challenges to put AI for general use is probably around return on investment. Many people deem AI as expensive, because the technology usually delivers value indirectly. It doesn’t generate revenue directly by making sales, but it helps employees boost productivity, and support other technologies for better performance.
For example, machine vision can improve precision on supply chain and minimise errors. It can reduce costs, but the results may not be reflected right away.
Similar for natural language processing (NLP), it can simplify many process and streamline analysis of languages for better communications between human and machines. All these processes can save cost and time, but the demonstration can be complex.
Hence, the economic benefits may not be as apparent as other technologies can immediately bring. Many background work done by AI usually goes unnoticed. The same set of traditional metrics for other technologies cannot directly apply to AI while justifying its success.
Another issue is about the confidence – sometimes it’s about the ease of use – on AI applications. There have been concerns over data security, believing that data feeding for machine training could be exposed.
The volume of such data could sometimes be overwhelming that stops many enterprises to proceed before any sound measures of security are in place. And the complicated structure may lead people think deployments are not straight forward.
Good AI should be easy to use, affordable and secure
Though AI may not directly address to the problems that enterprises are currently facing, the huge amount the data still requires tremendous computing power to handle. Since the 4G era, the increasing connectivity has led to data explosion.
For each person, we may need an average of 20 hard disks to hold our data in 2025, compared to just five now. The demand for AI would be even bigger to smartly handle such data flow for enterprises.
Huawei always believes in shared success. So, when developing feasible and sustainable AI solutions, the company emphasises on its ease of use, affordability and the security in the applications. The key is to turn artificial intelligence as a “essential power” for the fourth industrial revolution, making it the fundamental technology for the purpose of general applications across all sectors.
This is why it has developed the Atlas AI computing solution, covering services for all AI scenarios across devices, the edge, and the cloud.
The Atlas 900 AI cluster, for instance, provides data centres with powerful computing to supercharge computing-intensive research in astronomical exploration, weather forecasting, oil exploration, and gene research.
The Atlas 500 AI edge station, meanwhile, is especially designed for edge applications, meeting complex tasks in other scenarios such as video surveillance, transportation, community or campus, etc.
Atlas AI computing solution steps up to challenges
AI accelerates research and increases result accuracy to ultimately benefit everyone. On the edge side, Huawei is working closely with partners on a variety of AI solutions, such as AI-enabled inspection for transmission lines, smart customer service centres, medical image diagnosis, and manufacturing quality inspection.
Our goal is to make pervasive intelligence a reality with quick and simple AI. This is done by unifying the key components with the same architecture, so that everything can be done with just a few click. The simplicity helps bring the technology to more enterprises for their easy deployment and thus lower cost.
Huawei not only develops the relevant solutions, but also tests them within our own organisation, making sure the security standards are well met.
For example, it is using these systems along our production lines. This is also an attempt to address the mismatch between demand and supply of AI computing power. The current cost of such power is high as the size of user group remains small.
The potential, however, is tremendous. It is estimated that AI computing will meet 80% of the total computing requirements by 2025, and that the value of the global computing industry will climb from USD 1.5 trillion in 2018 to USD 2 trillion in 2023.
Huawei is bringing these innovative solutions to more industry partners to turn AI into a general-purpose technology accessible and affordable to more people and organisations. This indeed aligns with Huawei’s long-term mission to establish an ecosystem to include all partners and users for more advancement of this technology.
Talent cultivation for better growth of AI
The development of AI doesn’t stop here, but the lack of first-class talent can further hamper the progress. Some reports have revealed that there are some 300,000 AI engineers across the globe, which are far from enough for the technology to meaningfully expand in all industries.
Huawei has noticed the shortage and never stopped in cultivating these great people to drive the future to proceed.
The talent pool must be grown among younger folks, and these will be the basic developers for the technology in near future. That’s why Huawei has been investing heavily in aspiring talented scientists to flex their muscles.
Actions include the 100-million-euro investment over the next five years in the AI Ecosystem Program, helping industry organizations, 200,000 developers, 500 ISV partners, and 50 universities and research institutes to boost innovation.
Ultimately, Huawei is working hard to build a fully connected, intelligent world. This is the mission repeatedly communicated in the World Artificial Intelligence Conference, which took place in July, where experts from Huawei and the industry shared their insights into how AI can help bring us a better world.
To find out more, please visit Huawei website.