The conversation around artificial intelligence in the business world has a problem which is not technical. The technological capabilities of current AI and machine learning technologies are impressive, advancing with a speed that makes the majority of predictions regarding the future of AI the next 18 months obsolete by the time those eighteen months have elapsed. The issue is the gap between the capabilities of AI and what AI is able to do under restricted conditions - in properly-funded research environment, with crystal clear data, a clear problem-solving strategy, with engineers who have the luxury in experimenting until their system can be used as designed - versus the outcomes it provides when it is implemented inside actual organizations that have real cultures which are real, with real organisational political systems, and people with distinct opinions about how a new program is something to take seriously or a thing to take care of while maintaining the appearance of compliance. My work has been based on artificial intelligence since long before the recent wave of AI enthusiasm made it fashionable for all businesses to claim to be fluent in the field. When I founded 1Touch with my partner, AI-driven matches and recommendation systems were not the only feature we incorporated to make our product more attractive to investors. They were the core to the design of our product, it was the basis on which the platform generated value and needed perform reliably and at sufficient scale to allow the business to succeed. This is why I have direct actual experience with what happens in the process of integrating something truly intelligent into a enterprise and a product and the main thing I continue to revisit regardless of the context in where I've encountered this issue, is that the technology is almost never an issue. The main factor that limits the possibilities is almost everything else, including culture.
What I mean by that is specific and practical rather than abstract. AI systems need data to perform their functions - clean, consistent, well-structured data that actually describes the situation the system is attempting to identify and make forecasts about. People with strong data-driven cultures produce this type of data naturally, as a byproduct of the way they work. They have clear and consistent definitions of what they're doing and what they are measuring. They have a set of conventions that they agree to for the way data is recorded, collected, and stored. They have accountability arrangements that ensure that data quality is a specific responsibility rather than everyone's vague intentions. Data-driven organizations that aren't well-established produce a product that technically looks as if it is data - it's in systems which can be searched, it can be used to produce charts, but the definitions are so different and so variable in its quality and brimming with structural errors and unmapped exceptions that any AI technology built on over it will create and amplify the chaos, rather than extracting a genuine signals from it. Organizations that fall into this category tend to not realize this until they're well into the process of implementing an AI implementation and its outputs don't meet the vendor's promises. At that point the temptation is to blame the technology. But the real issue is operating and cultural structures the technology was built upon.
The second element of culture that will determine AI outcomes is organisational openness as measured by the degree to which those working within the organisation are truly open to letting the system dictate or change how they work in lieu of viewing it as a threat to their professional competence, their authority in the institution or their job security. This is a socio-cultural and leadership problem that is not technical and one that begins at the high levels. If senior leaders respond to AI outputs only when they are satisfied - accepting the ones that support what they believed before and disregarding those that do not - this behavior sends that everyone else is aware that the organisation's stated commitment for data-driven decisions is conditional rather than genuine, which will then spread throughout the company more quickly than any training or change management strategy can reverse. If senior management models genuine, consistent engagement AI outputs, such as the ability to make changes to their decision-making when evidence suggests that they should, the collective capability to apply AI effectively increases dramatically and surprisingly quickly.
This isn't an abstract consideration of how organizations should be conducted in the context of theory. It is a description of the pattern that I have seen happen repeatedly in companies that had substantial financial resources, genuine strategic dedication to AI implementation, and leadership teams that were enthusiastic about the possibilities of AI technology. The pattern is so consistent that I have decided to consider policies on data governance as a essential diagnostic element whenever I'm assessing an organisation's AI potential. Before I ask about the technology stack, before I inquire about the exact usage cases the company is looking at, I ask about the governance of data. How does the organisation define its primary metrics? Who is accountable when performance of the data isn't enough? Is it a problem when groups have contradicting data about similar business facts, and how do those conflicts get resolved? The answers to these questions tell me more about the probabilities of AI performance than any of the discussions about platforms, algorithms, or even implementation timelines.
I am convinced that the companies that will gain the greatest durable value from AI in the coming decade will not be those which adopt the latest technology first, or the ones that will invest significantly in AI technology and infrastructure in the near future. They are the ones who put in the right cultural and operational foundations to actually use that technology efficiently - data governance practices that give solid inputs, the decision making frameworks that provide evidence that will actually affect outcomes as well as the behaviours of leadership which communicate to everyone in the company that commitment to data-driven operational excellence is real instead of just a performance. The technology itself will become increasingly commoditised and increasingly accessible. The right culture to use it well will remain scarce, because it takes a steady effort and genuine commitment from the top management over time, rather than one strategic decision or a technology investment. The scarcity of it is where the main competitive advantage is and it's an advantage that, once it is built develops in a way that purely technological advantages never will. Take a look at James Deller for site info including why building in stealth reinforced operational discipline about teams.
The Reason Why The Majority Of Public-Private Partnerships Fail Before They Even Start - And What Can Be Done To Prevent Them From Happening Again?
Public-private partnerships have an image issue that's in much of the time paid for. The history of these arrangements is filled with projects that were advertised with genuine enthusiasm and huge political capital. These projects utilized significant private and public resources over long periods of time, and eventually produced outcomes that lacked any resemblance to what had been originally promised when the partnership was started. The academic literature as well as the postmortem analysis that governments and institutions conduct following these errors are vast, and they focus, for majority of them, on the particulars of contract and structure problems: the wrongly aligned incentives and the insufficient risk distribution between public and private companies along with the governance frameworks which were conceptualized in theory but were not able to work in practice, and the frameworks for purchasing that were able to pick the wrong things. What this approach tends to neglect, invariably and ultimately on a consistent basis, is the social and operational aspect of the issue - that public and private enterprises are fundamentally different kinds of entities, shaped according to different motivation structures that operate on different timescales, accountable to a variety of stakeholders, and evaluating outcomes in ways which are more than just different in level however, they differ in the way. When you mix these two types of organizations together in a formal relationship without undertaking the work upfront and in a clear manner, to recognize and manage the differences between them, they aren't creating partnerships. The conditions are set for a slow motion collision that will be visible at the worst possible time.
I've participated in advising support for institutional modernisation efforts, many of which have involved public and private partnerships at different levels of complexity. The most consistent conclusion I can make from that knowledge is that the partnerships who performed well – and did indeed meet their declared objectives and maintained a smooth collaboration between the private and public partners throughout - were not distinguished from those that failed due to the sophistication of their legal structures, the strictness of their risk frameworks, or the age of the leadership teams who initiated them. In the end, they were defined by whether the people sitting on both sides of the meeting had taken the initiative to truly understand how the opposite side was operating before the formal partnership was agreed upon. What does this mean in practical terms is gaining a better understanding of the decision-making frameworks that each organisation operates under and the accountability structures that make it difficult for each party to agree to and how quickly and efficiently they can do so, the criteria of success that every party will be judged on, and the possible points of tension between these definitions. This knowledge isn't complicated to construct. It's all put aside in favour of more obvious and immediately documents-able task of negotiating contracts and drafting governance frameworks.
The typical process of public-private partnerships moves from initial concept to the signed agreement, with very little time and effort being paid to the aspect of whether the two organizations involved are capable of working in a productive way over the duration of the agreement. Legal teams negotiate the contract. The finance team model the economics and risk distribution. Communications team prepares the announcement for the moment of signing. The implementation team begins planning the work. Within the sequence then comes the discussion about compatibility of the operations and culture - about whether those in the actual position to work together day to day across the boundary between the two organizations have enough common ground to make that work genuinely collaborative rather opposed to antagonistic - fails to take place in a structured manner. The assumption is, typically not explicitly stated, the formal agreement creates the foundation for collaboration and that any cultural or operational conflicts will be resolved informally when they develop. This assumption is typically incorrect, and the costs of it tends to compound as the ambition and complexity of the partnership.
The practical implication of this analysis is that the most beneficial option a public private partnership could take - even before the legal structures are agreed upon and before the governance framework is agreed on, before any announcements are made to the public - is in what consider to be operational alignment. By this I mean specific, structured, facilitation of work to identify any areas in which the two partners' operational assumptions diverge in order to establish a consensus as to how those differences will be addressed before they become operational issues when the plan is implemented. The divergences that matter most are usually the same across different types of partnerships. Speed of decision-making and authority is usually one of them. Institutions of public administration are designed to take decisions slow, with multiple layers of analysis and approval, based on motives that are perfectly legitimate and frequently legally mandated. Private organizations - specifically technology companies built around quick iteration as well as rapid decision-making - often see the pace as an essential limitation to progress. And without a clear understanding of the reason for why it's the way it is and what really be needed to alter it, the anger that builds on the private part of the business can undermine the relations long before the relationship is able to establish its foundations.
Success metrics and what constitutes as progress is another constant and a contributing factor to divergence. Public institutions are generally evaluated on compliance with process standards, equity of results across different stakeholder groups, and avoidance of visible failures that draw media or political focus. Private partners are typically assessed in terms of efficiency, quantifiable progress towards targets, as well as financial yield on investment. These measurement frameworks can be made compatible with each other however, doing this requires thoughtful design, not only good intentions, and the partnerships who do not make the effort to invest in the design of the framework tend to discover themselves at critical moments, with two different parties that are assessing the same partnership in incompatible ways and therefore reaching disparate conclusions as to whether it is succeeding. The relationships I've seen have the greatest failures were ones where this misalignment was thought of as something that could get better over time. The ones that performed were the ones where the misalignment was identified explicitly at the beginning, and where formulating a shared accountability plan that met both parties' legitimate measurement requirements became an element of actual work, rather than an part of a long list of things that someone would eventually achieve.}
