It takes more than technology to make businesses work. In fact, one of the most precious resources, besides the humans that run the organization, is capital to fund growth. In the areas at the edge of innovation, especially artificial intelligence, this capital is being used to prove that existing technology is capable of driving real business return as well as advance the state of the art to push the technology in ways not possible today.
It is no surprise, then, that major companies at the intersection of technology, manufacturing, and capital are seeing significant possibilities in artificial intelligence and investing both their time and resources to push the boundaries of AI’s capabilities. In a recent AI Today podcast interview, Sam Huang of BMW iVentures shared how the company is approaching AI and where it sees possibilities in investing in AI’s future.
Investment in AI
BMW iVentures focuses on investing in companies that are at the forefront of creating new and interesting technologies that meet existing business needs. Specifically, the venture fund focuses on investing in companies with technology that is directly or indirectly supportive of advancements in the automotive ecosystem. The fund actively invests in startups working on designing AI functionality for self-driving cars, intelligent systems that can help with services or might innovate an automotive industry process in some way, and various technologies related to transportation. The fund even invests in smart systems designed to help with finding automotive insurance, helping customers find their dream cars, and basically anything interesting that suits a need within the industry.
When evaluating how to invest in AI companies, the fund essentially sees two categories of AI relevant companies. The first category is general purpose AI companies, which are companies with interesting AI technology that can suit multiple needs and circumstances, whether relevant specifically to the automotive industry or not. These general focus AI companies are good investments because they solve general challenges faced by most companies looking to address AI needs and also enable the capabilities required by domain-specific solutions. These offerings are focused on adaptable technology that conforms to various AI needs.
The second category of AI investments are applied AI companies, which are companies providing industry- or domain-specific AI solutions relevant to specific application needs. Applied AI companies are some of the most diverse investment opportunities and can provide significant returns because they already know the problem that they are going to work on solving and have a focused market. Some of these solutions are focused on the automotive and transportation industry while others are focused on aspects of insurance or finance. Other domain-specific AI companies are applying their solutions to identify instances of fraud or simplifying processes that can be handled by software or hardware robots.
One off-shoot of these applied AI companies are startups that have not necessarily built their company to be identified as AI companies, but rather are leveraging AI technology to make their domain-specific solutions stronger, faster, more intelligent, and more efficient. Some of these startup companies are using machine learning and other aspects of AI to analyze big data to extract more value from information. These AI systems that can process large quantities of data and provide deeper insight and value beyond simple analytics of information.
Observing Patterns of AI Adoption
In the interview, Ms. Huang relates some interesting patterns she has observed with regards to AI adoption. The sorts of companies BMW iVentures is seeing and investing in are primarily using AI to focus on streamlining workflows, optimizing processes, and reducing overall costs. Since AI holds the ability to analyze complex datasets and identify data patterns very quickly, it can provide fast results and identify very specific needs or circumstances without necessarily relying on a team of people who need to try to process more than they can reliably count on. Already, AI has managed to identify trends that have helped to innovate the ways that companies do business, by providing customized customer interactions and identifying needs for clients. The biggest struggle with data, particularly in the automotive industry where the actual process of taking customer feedback and turning that into a future product can take several years, is ensuring that the data being referenced is still relevant. AI technologies aim to increase the speed with which data can be analyzed, which can get future designs started earlier than was previously possible.
The biggest concern with AI design and big data analytics is that the systems must be taught. Machine learning systems, and in particular those using supervised learning approaches, are dependent on clean, well-labeled data from reliable data sets to provide valuable and predictable learning results. While humans have the cognitive ability to notice patterns or anomalies in data, machine learning algorithms are entirely dependent upon the quality of data to create reliable models that can work with future, unknown data. This means that if the machine learning system is fed bad or incorrect data, the output model that is produced will be inherently flawed. This can be problematic in instances where data sources are limited or they have poorly collected data that is rife with errors, inconsistencies, inaccuracies, or missing data.
This need for reliable, clean data is especially important and relevant to autonomous vehicles and self-driving cars, which depend on those models to be able to successfully operate without human interaction. In the case of self-driving cars, the amount of required data that must be fed into the system before it can produce accurate outputs is extensive and can be incredibly time consuming to collect, clean, and label. This front-loaded need for quality training data slows down the development, testing and quality control efforts of self-driving systems. BMW iVentures sees the need to continue to invest in technologies and solutions to address data quality needs and ways to further optimize the training process. The net effect of improved machine learning training processes speeds up AI development as a whole and therefore accelerates all AI-relevant efforts.
Regardless of what industry you are in, the future will certainly be filled with AI technologies focused on implementing positive change and yielding better results. The capacity for systems to learn processes and analyze complex sources of data is one that will only continue to create better and more relevant results over time. Artificial intelligence is something that can be adapted to suit needs across every industry, whether it is in the form of bots helping organizations with their customer service needs, self-driving vehicles, or intelligent automation applications that provide the next level of productivity and performance required to advance to the next level of digital transformation. As an area of investment, AI continues to be rich with opportunity.
