“Although AI is already in use in thousands of companies around the world, most big opportunities have not yet been tapped.” This 2017 quote by Erik Brynjolfsson and Andrew McAfee, Director and Co-director of the MIT Initiative on the Digital Economy, is no less relevant in 2021. For almost three centuries, technological innovations have been the main drivers of economic growth.
It seems that currently, artificial intelligence, AI, is poised to have this kind of transformational impact. This technology is already widely known. Nevertheless, businesses are only just beginning to uncover the potential behind AI and its adjacent spheres. To better understand the economic, political, and organizational disruptions that developments in the AI sector are bringing, the M&A Community talked to Bruce Molloy, CEO of Springboard AI, a visionary, inventor, and a strategic consultant in the Artificial Intelligence and Machine Learning space.
The AI industry is booming now. Can we expect an ‘AI arms race’ among the top players?
Well for starters let’s talk about what AI is – and it’s very much a moving target. Years ago a system that could win at checkers or suggest spelling corrections was considered artificially intelligent. We’re now well beyond that. Capabilities once called AI are absorbed into regular software and are no longer called AI. A lot of recent success in AI has been the result of plentiful data in digital formats coupled with machine learning and its powerful, predictive capabilities. That being said, reinforcement learning, new classes of visual and audio recognition, adversarial networks and self-improving models, to name a few, will continue to advance AI.
Who are these top players?
There are a number of leaders in AI Software such as Facebook, Amazon, Microsoft, Google, and Apple (FAMGA), and yes, they are in an “arms race.” I believe Google has the best AI and will continue to be the leader, given its enormous global collection of content, its theoretical, scientific mindset (which is quite future-looking), and its stellar collection of talent.
Amazon, a more pragmatic contender, is a formidable number two. With its dominance in e-Commerce, it can mine all customer interactions while gaining powerful insights into every product and product class it carries. The Amazon “data hose” is constantly nourishing every aspect of their AI models. And Alexa is not insignificant. Alexa is able to mine not only the conversational exchanges with customers but also the nuances and emotional content in each customer’s voice. This collected data will help give rise to emotional intelligence and also provide initial building blocks for general artificial intelligence (GA).
The Microsoft strategy is quite different. Given that it has a foothold in nearly every major Corporation in the form of word Excel, SQL, PowerPoint, etc., Microsoft can incrementally add whole new sets of AI features and enjoy immediate adoption. Like Google and Amazon, Microsoft has access to tremendous amounts of data; however, in Microsoft’s case it is corporate data, not internet, consumer data. Microsoft’s strategy could be thought of as DIY AI which will help front-run the democratization of AI. They are also a strong contender.
Facebook is good at what they do, and Apple is playing catchup. There are many other challengers – and any number of new disruptors who were “born AI.” In AI hardware, chip manufacturer Nvidia is a clear leader with challengers like Intel and Graphcore.
What countries have the potential to enter the major league?
To determine who might be leading this race, one could simply count the number of patent applications filed by country. This past fall, China led in the number of applications filed, followed by the US, South Korea, India, and Europe (as a group). China’s 4,600 applications represented twice as many applications as the other four combined. By this measure, China is clearly ahead.
Are there any other parts to this story?
Surely. Innovation, culture, national support, and financial resources (think VCs) are also key determinants. Essential is the risk-taking, creativity, and entrepreneurial cross-fertilization coming from tech hubs like the Silicon Valley and other similar environments: Tel Aviv, Berlin, Lisbon, Austin, etc. The venture capital community is also critical as it provides coddling and enormous financial support to these startups. Breakthroughs require the freedom to experiment and fail – and to succeed.
An example of experimentation and disruption is found in Geoffrey Hinton. Hinton, a luminary now at Google, was one of the seminal pioneers in neural nets and deep learning. When he first came on the scene, AI was shaped by the austerity of logic and math in the form of expert systems and symbolic computing. Hinton thumbed his nose at this and fashioned a richer and messier approach inspired by biology and human neurons. It worked, and it is still working today quite well.
Time will tell who will win the AI arms race, a race which should more properly be thought of as multiple races given the many facets of AI (cybersecurity, military, healthcare, marketing, finance, autonomous driving, robotics, and the list goes on.) The lead may change multiple times in each race. While the US has a remarkable entrepreneurial spirit coupled with immense VC backing, China is a nation obsessively focused on AI and is providing outsized resources through government investment.
I think the US can win many of these races if it puts together a national “moon shot” initiative as it has done in the past. But if the US approach continues to be entirely market-based and piecemeal, without the resources and coordination of the US government, it’s a coin toss who wins.
Endnote: Just as we’re getting a handle on the odds of these races, quantum computing will begin commercial deployment in 2023-2026. Then a whole new set of races begin – stay tuned for “Quantum Machine Learning.”
AI is a highly complex, technical topic. From your point of view, what should any business leader know about AI?
AI typically combines computer engineering, data science, logic, multi-variate calculus, and statistics, along with domain knowledge and a good amount of data. Also, implementing AI requires industrial-strength scaling and deployment analogous to the factory that scales up models created in the laboratory.
This being said, business leaders should not be intimidated at all; they can understand what AI can do in straightforward ways, including the opportunities it creates, the problems it solves, the risks it poses, and where it is going. Each of these four areas can be understood conceptually, intuitively, and with accessible examples. Business leaders need to have a confident grasp of the subject, not feel intimidated, and importantly, not get drawn into the weeds.
Here’s what Bruce Molloy recommends for any business leader.
Rapid AI immersion
You should get up to speed on AI quickly. The C-Suite and board should consider developing a conceptual framework for thinking about AI, and good examples to support their understanding. Workshops and training should be given at all levels. DIY tools such as Microsoft’s Azure or a myriad of others can be made available to all who want them, like the way all employees were given Word, PowerPoint, and Excel when they first came out.
One effective approach for senior managers and/or the board to get down the learning curve would be a 3-hour workshop, split into two sections. The first section would be an overview with case studies and examples. The second section would be a group exercise to evaluate top priority use cases for the business. These use cases can then be stack ranked according to:
- Impact
- Data availability
- Resources required
- ROI
The top use case becomes the basis for a 90-day “AI Sprint” with “proof of concept” and “proof of value” being the goals. Ideally one should strive for 10x-ers, although getting part way there should be a success. This is a way to get senior management and the board involved, track results, and create a starting point for larger implementations.
Dive in deep and take a good look around
Find out what your competitors are doing, find out what your industry is doing, find out what other industries are doing; then bring these learnings back into your own company.
Get AI connected throughout the organization, not in silos
I had one CTO proudly tell me he had cloistered 11 PhDs in a lab up in Rhode Island. Do not take this approach. Make sure AI practitioners are embedded within business units and vice versa. Get AI into the fabric of the organization where it can be informed by business domain expertise.
Realize there are no experts who know it all
There are too many dimensions, too many industries, too many applications. Find meta-experts who can create “quick assemble” teams for each application.
Give serious thought to your AI data strategy
Are you collecting the right data? Simply having enormous data warehouses and data lakes does not make this data valuable. This data could be like the old boxes of stuff we all have accumulating in our attics.
Should you “instrument” or create new data collection capabilities in the areas of operations, sales and the marketplace in order to get more of the right data? Are you monitoring your data for quality, bias, regulatory compliance, etc.? Do you have data governance in place? Perhaps most importantly, are you collecting data with a strategic mindset to the future, such that even if you do not use the data right now, it could be highly valuable in the future for various AI models including prediction?
Always keep an eye to the future
I promise you in two years the world of AI will look much different than it does now. Certain projects that currently require teams of top scientists, will be realized simply through plug-and play-platforms and “no-code AI.” Some essential data will be made available on data exchanges, and there will be a much larger team of engineers for hire who can pull all this together. Processing power using quantum computers will increase at a “doubly-exponential” rate according to Neven’s law, way beyond the speed of change predicted by Moore’s doubling law. Value creation will continue to shift from engineering to innovation and strategy as whole new types of products and business models emerge. Be forever thinking new products and new business models enabled by AI.
