This is the fourth and last round of our series “Bridging digital capability gaps: an AI perspective”. This episode takes a different format: readers and clients are invited in the driving seat to put the questions in.
Q: We have an exciting conversation ahead of us on the topic of AI ecosystems. So the first question, naturally is what is an AI ecosystem ?
To start off, there are lots of books and even more many excellent chapters and articles written about business and digital ecosystems. So this is not a new concept. Instead, it is the rapid progress of digital technologies that have propelled the term to a whole new level. And AI ecosystems can be defined as groups of interdependent entities aiming at delivering AI products and services.
Yet, as the digital economy requires value creation as the main driver of digital ecosystems, similarly participants of AI ecosystems are not entering a kind of supply chain but are orchestrators or active contributors coming with new products and business models, intangible assets such as knowledge, ideas and relationships, and then helping to pursue meaningful impact. And this last bit is important because it is generally accepted that the real power of AI can only be unleashed by focusing on AI that works for all, and can demonstrate socio-economic and environmental benefits. As a result, this discussion will seek to ferret out and present some of those initiatives and programs that can bring AI to new heights.
Q: What have governments, businesses and institutions done to encourage budding entrepreneurs and promising startups ?
National AI Strategies have emerged, which have created an environment where AI ecosystems can thrive. As an example, the report Canada’s AI ecosystems produced by the University of Toronto showcases the strength and success of the Strategy put in place that contributed to making Canada one of the leading global players in AI – the cities of Toronto and Montreal having the major concentrations of AI firms. In the same vein, the Autorité des Marchés Financiers plays an active role in the thriving Québec fintech ecosystem including support of innovative firms as well as exploring new technologies in its fintech lab.
A variety of initiatives and programs aim to support innovating companies and early-stage startups to scale up, providing guidance and advisory from experts, office support and a community of learning. SMEs and startups will need to be part of centres of excellence, which are comprised of research & training institutions, incubators & accelerators, financial & funding actors, and also large companies.
Also, initiatives should aim to foster innovation and collaboration, in order to develop and accelerate the adoption of AI. In this regard, Microsoft introduced in February 2010, the Turing Natural Language Generation through the AI at Scale initiative, a platform made available to developers to infuse AI into their applications.
In the UK, the Knowledge Transfer Partnership (KTP) is run by the UKRI (Research & Innovation). It’s a scheme whereby businesses get a shot to innovate, develop, grow and become more productive by employing academic expertise that they do not have in-house. It goes further to provide financial support to graduates who will work on innovation projects for a period of 12 months up to 36 months in those businesses.
Another example of successful initiatives is the Digital Sandbox Ecosystem launched by the City of London and the Financial Conduct Authority. It provided a digital testing environment for innovating companies to come up with solutions to problems caused by Covid-19, including malevolent emails, basic access to finance, and supporting the elderly.
Therefore, whether it’s offering synthetic data or connecting with peers, there are a number of initiatives and programs that can be implemented.
Q: When it comes to medium-to-large companies they seem to have more resources to deploy AI systems. Yet, smaller tech firms including Insuretech, Regtech and Fintech companies have challenged these incumbents.
Indeed, and they have done so for the better because overall, innovation has accelerated as a result. This trend can also be seen with intrapreneurship ventures where companies put together a dedicated team to work on innovative AI projects. Yet, under certain conditions defined by the alliances and partnerships, we also have room for those startups to contribute.
Let’s start with the Telecoms industry which has recently experienced rapid developments with the advances of 5G and IoT devices. AI has become essential to address the increasing complexity that came with virtualization and cloud computing. Some of the AI use cases include congestion & fault predictions, performance monitoring & optimization, and automated resolution of trouble tickets. There is also the need to clean and categorize data before it can be used to train ML systems. And these areas could be some entry points for startups.
In healthcare, data is notoriously expensive to collect and label. AI techniques can help to discover new drugs, improve quality of life and assist in robotic surgery. The low-hanging fruits will be virtual health assistants, customer service chatbots, claims handling and settlement. This is even more important given the great prominence of the wellness industry as well as the new opportunities offered by the value chain and distribution channels in the life & health Insurance industry.
In the financial sector (Banking & Insurance), credit risk modelling is relatively low among AI use cases because of the issues around explainability of the models. As a result, “hybrid” or “middle-of-the-road” solutions are currently favoured over complex ML methods.
In the realm of financial /wealth advisory, robot advice has appealed for some time now to the new generation. Instead of a 2-hour face-to-face meeting, the preliminary work of a robot could take much less than 20-25 mins to make a proposal to the user of the application. At the turn of 2017, while asset managers enjoyed exceptional results after plodding performances from previous years, they had to turn their attention to services of high-end value, especially when pressures on margins were becoming ever more increasing, partly because of the advances in digital and analytics. And with many APIs being able to collect data from various sources, startups can chip in to provide some automated solutions.
Moreover, traditional Financial Planning & Analysis (FP&A) are being upskilled with automation and predictive analysis.
There is also a large data science subfield which may not be human interfaces or AI-powered design as such but more analytic-driven.
In Manufacturing, the potentials are huge for AI to transform the factory of the future, with opportunities in production lines, inventory management and detection of product defects.
As for top AI use cases, a very good reference is the latest report from McKinsey Global Survey on The state of AI in 2021. Ultimately, for a range of sectors & industries, it is AI applied to reduce errors, speed up operations and optimize processes that will grow faster, and directly impact the bottom line.
Q: As AI companies become more sophisticated, what currently defines the next AI frontiers, in terms of future developments ?
It certainly depends on specific fields and applications of AI in various sectors and industries. But active research is underway to determine if we can do more with less data, given the deluge of it. Yet, as AI has also increased dramatically thanks to the advances in computational power and the prowess of DL algorithms, a common denominator will be to find out if we can have more “intelligent” systems by making significant improvements to one or more of these technological breakthroughs.
Besides, recent years have seen the hype and development of innovations such as IoT, Blockchain, digital twins, quantum computing. With global topics ranging from decentralized finance to pandemics to net-zero emissions, these transformational technologies, when combined with AI will help to unlock even more value from data, which is the glue that holds them together. They promise breakthroughs in entire industries from aerospace to health to synthetic biology to advanced materials to food, water and agriculture in order to solve many of the world’s most pressing problems. And the UN SDGs provide an excellent framework for assessing the impact of AI on various sectors.
Commonly referred as DeepTech, are those advanced research in technologies and products that will be launched in 5, 10 years or more. DeepTech also faces other hurdles including expensive commercialization and IP that only large companies can afford. Yet again, almost all of DeepTech ventures are linked to universities and research institutions in order to foster – according to Goal 17 of the SDGs – partnerships for collective action across the public, private and social sectors.
Q: Now it appears that the most developed countries would have better growth and increases in productivity in the future as a result of their lead in AI. What can be done ?
As AI is pervading all sectors and industries, many governments, the world over have by now identified its potential and benefits. And it is very likely that the divide will keep increasing between the main players and those lagging behind.
At the core of the strategic plans designed by nations is often to adopt a portfolio approach considering near-term requirements and long-term objectives. Recommendations suggest to focus their efforts on enabling the foundational AI infrastructure (the stack of hardware and software necessary to reap the full benefits of AI technologies). From there, entry costs are much lower as it opens the door for freelance work or the gig economy. Indeed, it is said that AI may be more of a services business than initially thought. Pre-trained models and libraries are easily accessible to software engineers to implement AI systems. Whereas, historically a PhD plus 5 years experience in the field was required to work in AI, today willingness to learn should suffice.
Data Science student challenge «Flights» was a competition that took place over the few weeks ending to January 2022. It was jointly organized by the Data Science et Processus Industriels (Université Mohammed VI Polytechnique) and the Data Science Institute (Institut National Polytechnique Houphouet-Boigny). Flights, which brought together over 100 students from some 10 countries proposed to build supervised learning models that will predict the exact number of passengers on a plane – all with real data. Such initiatives have a goal to encourage innovators, and train future decision-makers in Africa.
In the longer term, the presence of Angel or crowdfunding networks or higher returns for VC will ultimately yield a flourishing environment. Reports show that tech startups are growing by leaps and bounds in Africa, faster than in the rest of the world. The African Development Bank’s (AfDB) has launched the Innovation & Entrepreneurship Lab which is an initiative with the goal of supporting the youth employment and entrepreneurship ecosystem. Africa is known for being a fertile ground for AI to thrive. However, the continent has only a small bunch of unicorns or zebras. As a result, governments, many leading African corporations and development institutions have sought to mitigate the inherent risks and to enable the environment for ease of doing business.
Q: Time to wrap up ?
Data-driven technologies like AI hold the promise to transform many sectors and industries, our lives and the society we live in. Yet, as scrutiny increases for explainability of black-box models, enterprises would choose, understandably to shift their focus towards more straightforward AI use cases.
In November-December 2021, we attended the Responsible AI series organized by the IET. The live talks with Q&A addressed issues of concerns not only from the technological perspective but also through the societal lens.
From navigating the rules of the game, to crafting inclusive business models to building capabilities & human capital, there are significant challenges but real chances to expand economic opportunities for AI ecosystems. The outcome: a shared value for both business and society, each having a multiplier effect on development.
Alongside large players and orchestrators, smaller contributors can play a major role within AI ecosystems. That is by developing products and services that people use, by creating innovating technologies or platforms that become key building blocks for future impactful systems.