AI or artificial intelligence – as an abstract concept – conjures up multiple images, in part because it has been so loved by science fiction writers and film makers over many years. It certainly wasn’t all good, was it? In my mind, straight off the bat, there’s the unnerving and ultimately misfiring HAL from Stanley Kubrik’s ‘2001: A Space Odyssey’. Or there’s ‘The Terminator’and the dystopian future where the machines have taken control of everything, humans rendered obsolete and no longer needed. And then there are the many robot adaptations and androids, machines with human characteristics or indeed, in human form.
Creative minds certainly like that interplay of humans and robots/machines that might eventually do the things that us humans do. There’s also something of a debate about what intelligence or human-style thinking really is and where emotion comes in. It can get very philosophical.
In the first part of the 21st century, AI has kind of come of age – but we’re still in the early days of its development. Definitions vary but the realities of AI in 2021 are a little more prosaic than the outlandish products of the imaginations of science fiction writers. IBM (of all people they should know) define it as ‘leveraging computers and machines to mimic the problem-solving and decision-making capabilities of the human mind’.
At its simplest form, according to IBM, artificial intelligence is a field which combines computer science and robust datasets, to enable problem-solving. IBM also says it encompasses sub-fields of machine learning (ML) and deep learning and these disciplines are ‘comprised of AI algorithms which seek to create expert systems which make predictions or classifications based on input data’. ‘Deep learning’ comprises so-called neural networks, or layers of inputs and outputs, a kind of ‘scalable machine learning’ but machine learning is the bedrock of all of this. ML is based on the premise that systems can be designed to ‘learn’ from data, identify patterns and make decisions with minimal human intervention. Very clever algorithms (written by humans, who set the parameters for ML decision making) have started this particular ball rolling and big data sets – like those generated by connected cars – are the abundant raw material that clever chips can process increasingly quickly.
Advanced high speed computing capabilities available today enable or facilitate many machine learning applications – and we’re seeing more. It is becoming less ‘exceptional’ and more ‘mainstream’. Digital processes that create big datasets act as a facilitator. Large volumes of data can be quickly processed in real-time for solutions described variously as smart or intelligent.
The automotive industry is emerging as a major source of AI and machine learning. The importance of artificial intelligence (AI) to the automotive industry over the coming decade cannot be overstated. Facing the long-term existential threats of sustainability, overcapacity, and the prospect of decreasing volume due to the challenge of shared mobility, automotive players must harness AI’s potential. The greatest potential lies in the abundance of data that auto suppliers and automakers amass and do not currently use effectively.
Data volume will only continue to grow as autonomous, software-defined, and connected vehicle functions increase in number and scope.
Data science and machine learning (ML) are designed to quickly assimilate large volumes of data, understand what it means, and promptly apply the insights that emerge.
Moreover, the cash conservation and cost-cutting of moonshot projects brought about by the pandemic means that some of the threats (like autonomy and shared mobility) have temporarily abated. Hence now, more than ever, is the time to embrace AI in the automotive value chain.
AI has use cases beyond autonomous vehicles
Autonomous vehicles (AVs) are the most public-facing application of AI in the automotive sector. AI chips, computer vision, and ML are the key AI technologies associated with self-driving. However, AI is important across the whole value chain. Upstream (tier-1, 2, and 3 suppliers and automakers) benefits from computer vision and smart robots alongside data science and ML to streamline production, while downstream (sales and the increasingly important aftermarket) profits from conversational platforms and context-aware systems alongside data science and ML.
More importantly, AI plays a crucial role in closing the feedback loop between upstream and downstream by incorporating sale and post-sale vehicle data into predictive modelling, regulating production more closely to demand. Automakers can thus operate in an agile relationship with real-world events, which is necessary to mitigate crises like the pandemic and the automotive chip shortage, in addition to the threat from mobility challengers. Automakers and suppliers are finally realizing that they are far behind the software giants and are rightly wary of handing over value-add opportunities. Developing AI capabilities is now central to automakers’ future profitability and survival.
‘Digital twins’ technology
Digital twins use a combination of IoT sensors, real-time analytics, and ML to create a virtual simulation of an asset, factory, or supply chain. Constantly updated with new data gathered at the edge, the use of data science and ML in digital twins helps create a virtuous feedback cycle that enables earlier detection and prevention of problems causing inefficiencies. Furthermore, when the physical environment is modified based on such insights, new information is subsequently produced for the twin to assimilate and refine.
For automotive manufacturers, the end-to-end data picture provided thus could help rebalance supply chains proactively and quickly in the face of rapidly changing situations. Therefore, production can be transformed from reactive and siloed activities to a holistic, iterative, and agile process. AI can therefore enable automakers to operate in a much closer relationship with real-world events, which is exactly what needs to happen to survive and adapt successfully to future crises.
Smart cities overlap
The use of AI in automotive manufacturing will increasingly overlap with the development of sustainable smart cities. 5G connectivity will provide a bedrock of low-latency communication from vehicle-to-vehicle (V2V) and eventually vehicle-to-everything (V2X), which opens up a whole range of AI use cases. From a sustainability perspective, the prediction of road demand and centralized traffic management will benefit from AI, improving travel efficiency and lowering vehicle energy consumption. Further AI adoption will occur in fleet management and real-time vehicle routing by mobility providers and the enablement of ambient commerce in infotainment systems via smart infrastructure interaction.
The development of AI is naturally crucial to the potential success of Level 4 and 5 AVs, which will be heavily scrutinized by regulatory authorities before being taken up by the public. AI chips, computer vision, LiDAR, and edge compute power are the key technologies that are being rapidly developed for safe and reliable AVs to meet this most acute challenge. A low rate of failure is not palatable or acceptable when scaled up to hundreds of thousands and eventually millions of vehicles.
How AI can boost car company profits
AI can play an important role in arresting the diminishing bottom lines of automakers. In the shorter term, it will be key to make use of the increasingly granular levels of data available on vehicles, parts usage, and driving habits. ML and data science are vital tools that enable flexible demand planning strategies, thus maximizing cost reduction.
In the long term, as ownership and vehicle volume decrease, automakers will have to build entirely to demand, perhaps, in the most advanced circumstances, becoming captive suppliers to fleet operators. This will require smarter production methods and factories to reduce costs and maintain a viable profit margin. The use of AI to dictate supply chain management alongside the use of smart robots in factories will go a long way in reducing long-term costs despite the initial capital expenditure required to implement the technology.
Revenue streams are likely to come increasingly from value-added services rather than the traditional streams of vehicle sales and aftermarket part replacement. The biggest prospect is generating income by offering wirelessly delivered services, features, and upgrades made possible by the connected car. There may also be the possibility of earning commission on third-party purchases made through vehicle infotainment systems. Therefore, the AI systems behind the personalization trends of other sectors can undoubtedly be applied to the car market and will be crucial to respond to the threats of decreased volume and profitability. Automakers need to strike a balance between using the large technology companies’ superior AI and big data capabilities without totally ceding the potential value-add revenue available.
The reason for AI’s increasing importance in mitigating these challenges is due to the increasing homogenization of mobility vehicles. This means that, in the future, consumers will grow accustomed to prioritizing vehicle function over form. They won’t pick the best vehicle, but the best service and AI will help deliver the best services. Manufacturers and fleet managers that deploy AI most effectively to operate in the closest conjunction with customers’ preferred (and fluctuating) mobility demands will have the edge. AI is, therefore, a crucial tool to capitalize on this hyper-premiumization of function over form.
See GlobalData report: AI in Automotive