April 9, 2016 Leave a comment
After several decades of severe volatility in climatology across the fields involved with artificial intelligence (AI), we’ve finally breached the tipping point towards sustainability, which may also represent the true beginnings for a sustainable planet and humanity.
Recent investment in AI is primarily due to the formation of viable components in applied R&D that came together through a combination of purposeful engineering and serendipity, resulting in a wide variety of revolutionary functionality. However, since investment spikes also typically reflect reactionary herding, asset allocation mandates, monetary policy, and opaque strategic interests among other factors, caution is warranted.
The following considerations are offered as observations from my perch as an architect and founder who has been dealing with many dozens of management teams over the last few years. The order of priority will not be the same for each organization, though in practice are usually similar within industries.
Risk Management and Crisis Prevention
The nature of AI when combined with computer networking and interconnected emerging technology such as cryptography, 3D printing, biotech and nanotech represents perhaps the most significant risk and opportunity in history.
While the global warnings on AI are premature, often inaccurate, and appear to be a battle for control, catastrophic risk for individual companies is considerable. For most organizations the risk should be manageable, though not with traditional strategies and tactics. That is to say that AI within the overall environment requires aggressive behavioral change outside comfort zones.
Recent examples of multi-billion dollar investments in AI include Google, IBM, and Toyota, though multi-million USD investments now number in the thousands if we include internal investments and venturing. To be sure much of this investment is reactionary and wasteful, but the nature of the technology only requires a small fraction of the functionality to prove successful, which can be decisive in some markets.
For appreciation of the sea change, common functions employed today were deemed futuristic and decades in the future just three or four years ago. So it’s not surprising that a majority of senior management teams we’ve engaged in the last two years confirm that AI is among their highest priorities, though I must say some are still moving too slow. We’ve observed a wide range of actions from window dressing for Wall Street to confusing to brilliant.
The highest return on investment possible is prevention of catastrophic events, whether an industrial accident, lone wolf bad actors, systemic fraud, or disruption leading to displacement or irrelevance. Preventable losses in the tens of billions in single organizations have become common. Smaller events that require a similar core design to prevent or mitigate are the norm rather than the exception, but are often nonetheless career ending in hindsight, and can be fatal to all but the most capitalized companies. We’ve experienced several multi-billion dollar events in former management teams that likely could have been prevented if they had moved more quickly, including unfortunately loss of lives, which is what gets me up at 3am.
An exponential surge in training is underway in machine learning (ML) along with substantial funding in tools, so we can expect the cost of more common technical skills will begin to subside, while other challenges will escalate.
“In their struggle against the powers of the world around them their ﬁrst weapon was magic, the earliest fore-runner of the technology of to-day. Their reliance on magic was, as we suppose, derived from their overvaluation of their own intellectual operations, from their belief in the ‘omnipotence of thoughts’, which, incidentally, we come upon again in our obsessional neurotic patients.’’ — Sigmund Freud, 1932.
The challenge is that the magic of the previous century has evolved and matured by necessity from the efforts of many well meaning scientists, but some of the magicians on stage still suffer from neurosis. Technology is evolving much faster than humans or organizations.
Examples of talent issues commonly found in our communications:
Due to long AI winters followed by the recent tipping point in viability, the number of individuals with extensive experience is very small, and most are at a few tech companies attempting to displace other industries.
Industry-specific expertise beyond search and robotics is rare and very specialized with little understanding of enterprise-wide potential.
An exceptional level of caution is warranted on conflicts in AI counsel due to competency and pre-existing alliances.
Despite efforts to exploit emerging opportunity, ability to think strategically in AI systems appears to be almost non-existent.
CTOs may win some key battles in tactical applications, but CEOs must win the wars with organizational AI systems.
The talent war for the top tier in AI is so severe with such serious implications that hundreds of millions USD have been invested for key individuals. Of course very few organizations can compete in talent auctions, which is one reason why the Kyield OS is so important. We automate many AI functions that will be common in organizations and their networks for the foreseeable future while also making deep dive custom algorithmics simpler and more relevant.
Not only is AI a classic case of ‘offense is the best defense’, when designed and executed well to enhance knowledge workers and customers, the embedded intelligence with prescriptive analytics can accelerate discovery, uncover previously unknown opportunities, providing historically rare potential for new businesses, spin outs, joint ventures and other types of partnering. Managed well, this is precisely what many companies and national economies need.
Impacting every part of distributed organizations, the importance of architecture cannot be overstated as it will influence and in many instances determine outcomes in the distributed network environment. AI is a continuous process, not a one-off project, so it requires pivotal thinking from two decades of fast fail lean innovation that our lab helped pioneer. Key considerations in architecture we incorporated in the Kyield OS include but are not limited to the following:
Optimizing the Internet of Entities
Governance, compliance, and data quality
Accelerated discovery and innovation
Security and privacy
Ownership and control of data
Audits and reporting
A priority outcome for most organizations in competitive environments, productivity improvement is increasingly derived from optimizing embedded intelligence, which is also desperately needed to improve the macro global economic situation. A large gap remains in most AI strategies with respect to enterprise-wide productivity, which represents the foundation of recurring value to organizations and society, regardless of the specific task of each knowledge worker and organization.
While cultural challenges and defensive efforts are common obstacles to any productivity improvement, strong leadership has proven the ability to triumph. Internal and external consultants and advisors can help, particularly given the steep learning curve in AI; just be cautious on unhealthy relationships that may have interests directly opposed to the client organization, as conflicts are pervasive and tactics are sophisticated.
Just when we thought trust couldn’t become more important, it seems to dominate life on earth. We’ve come across quite a few trust related issues in our AI voyage. A few examples that come to mind:
Intellectual property: Trust is a two-way street, particularly when it comes to intellectual assets, so upfront mutual protection is a necessary evil and serves as the first formal step in establishing a trustworthy relationship, without which the other party must presume the worst of intentions. Once the Kyield OS is installed with partners this problem is effectively eliminated with smart contracts and digital currency based on internal dynamics and verified intelligence (aka evidence).
Fear of displacement: Since AI is new for most, suffice to say that fear is omnipresent and must be dealt with in a transparent and intelligent manner. At the knowledge worker level we overcome the problem with transparency, which makes it obvious that the Kyield OS is likely their strongest ally.
Modeling: While motivation to change is often needed from external sources such as regulatory or competition, it’s probably not a good idea to trust a company that has the capability, desire, culture and incentive to displace customers. Another problem to avoid at the confluence of networked computing and AI is lock-in from technology or talent, including service models. Beware the overfunded offering that attempts to buy adoption and/or over-reliance on marketing hype.
Authenticity: Apart from the serious structural economic problems caused by copying or theft of intellectual work, consider the trustworthiness of those who would do so and how much know-how is withheld because of this problem. Authenticity is especially important in this field due to the length of time required to understand the breadth and depth of implications across the organization and network economy.
Given the strategic implications to organizations, AI should be a top priority led by senior management. However, since supply chains face similar challenges with AI, traditional methods and channels to technology adoption may not necessarily serve organizations well, and in some cases may be high risk. Whether for strategic intent, financial return, operational necessity or any combination thereof, investing well in AI is not a trivial undertaking. Integrity, experience, knowledge and freedom from conflicts are therefore critical in choosing partners and investments.
About the author
Mark Montgomery is the founder and CEO of Kyield, which is based on two decades of self-funded R&D. The Kyield OS is designed around his patented AI system, which tailors data to each entity in the digital network environment.