Optimize your business with Causal AI

The business has no Artificial Intelligence (AI) capabilities, though there may be a well-established digital infrastructure that can manage data across business functions. Solutions to the businesses’ highest value problems depend entirely on the intelligence of the human workforce.

One in two enterprises are currently still at L0. They fail to realise the value of their data. There is firmwide hunger for ML resources, but no way to supply the resource. Business functions burn fuel improvising solutions instead. Organizations at this level are at a disadvantage as compared to competitors who have begun their journey to AI adoption.

Machine learning (ML)  — based on algorithms that learn from data and are not reliant on rules — is used in at least some business functions. A data science team is required to explore, build, test, scale and maintain models in production, with minimal automation of the end-to-end process.

The business can begin to participate in some of the benefits that ML delivers, in terms of process efficiency, superior forecasting and intelligence gathering. However at L1, data scientists are inefficient and often reactive, responding to requests from business functions seeking insights. The business is playing catch up with AI leaders, but it’s attempting to catch up from the slow lane.

The business uses ML collaboration platforms or software that automates ML solutions (autoML). Data scientists are more efficient and focussed on higher value problems. ML is democratised throughout the enterprise. The interpretation of ML models and their application to business decisions remain manual tasks.

65-85% of data scientists’ time is liberated by AutoML. New and unexpected value is unlocked from data, as the data team is able to focus on more aspirational projects and as other business functions benefit from direct access to advanced analytics. 

However, the technology is fundamentally limited, and only delivers results in certain narrow contexts. ML algorithms fail to adapt to new problem domains or situations, and overfit to the past on time-series data. Models are “black boxes” that no one can comprehend or interact with, and which do not generate accessible insight. Algorithms greedily require huge amounts of data, and energy, to do anything useful. At L2, the business has probably reached a bottleneck created by the inherent limitations of conventional ML.

Enterprise software automates Causal AI, which supersedes conventional ML and is a step towards true AI. Causal AI makes adaptable predictions in dynamic, complex environments, has a scientific understanding of markets, provides human-friendly explanations, designs optimal business plans and can “imagine” business improvements that are unconstrained by historical data.

Causal AI retains the benefits of AutoML in terms of automation of the data science pipeline, only the core technology is far more powerful. It brings more ambitious, higher-value projects into easy reach, and expands the frontier of data science applications across the enterprise. 

At L3, organisations enjoy deeper levels of human-machine collaboration. Domain experts can seamlessly share knowledge with AI systems via a shared language of causality. Business users can use AI systems to generate optimal plans and interventions, and can partner with AI systems that can now answer “why” and “what if” questions, form hypotheses, validate and disprove them.

AI systems are capable of autonomously optimising for abstract strategic goals and KPIs (autoKPI), and manage significant aspects of the enterprise. Enterprise software can coordinate multiple predictive and causal models, understand business context and manage uncertainty, enabling AI systems to solve complex business challenges.

L4 businesses are on the road to becoming self-sustaining and self-improving systems. Business improvements are accelerating at the pace of the movement of data. Leaner, more intelligent operations unlock new growth opportunities and win market share. 

Humans and machines have symbiotic relationships that involve significant mutual understanding and trust, and that facilitate thought partnership and ideation. Humans maintain control over corporate strategy and some operational targets, but operationalisation is increasingly owned by AI.

The enterprise is controlled by, or has become, an Artificial General Intelligence (AGI) — a general-purpose technology at least as intellectually advanced as humans. Super-intelligent strategy and maximally lean operations lead to an exponential increase in growth. Humans are liberated from work.

AGI is beyond the technological horizon for now. Expert opinion varies widely when it comes to forecasting whether we will succeed in engineering AGI, how we might plausibly develop it, and the time-frame for getting there. If reached, we envision AGI having a dramatic impact on businesses.  There may be a “winner-takes-all” effect for companies that first reach this level. We envisage businesses that reach L5 doubling their enterprise value on a monthly or even weekly basis. The current economic development model may change beyond recognition.