The first generation of the internet focused on connecting businesses and people, and the second wave centred on social interactions. Now we are entering the third generation, called the “Internet of Convergence” or the “Internet of Value”. This internet will be mostly boosted by the confluence of different technological advances that will allow us to design disruptive and complex systems gifted with ubiquity, hyperconnection, decentralization and intelligence.

Throughout this generation many industries, including banking and FinTech, will create new services, products and platforms based on the merging of exponential technologies: machine learning (and especially reinforcement learning), cryptocurrencies and blockchain, mobile phones, the internet of things, quantum computing, virtual reality (VR), augmented reality (AR), APIs, digital identity, decentralized and autonomous networks, cybersecurity and, of course, the data economy.

This new technological environment will lead to big opportunities in the financial industry. One of the most controversial effects, according to the futurists, would be “financial singularity” – an event that would change irreversibly the investment industry within 20 years.

The Artificial Intelligence Super Explosion

Before explaining what financial singularity is, you need to understand an underlying and more general and widespread theory: technological singularity. This concept refers to a hypothetical and crucial event in human evolution when, as a result of the advances in technology and the processing capacities of the new digital processors, for the first time in history, engineers will build artificial intelligence (AI) more intelligent than the average human brain.

This advanced AI could start a chain of self-improvement cycles, with new and more intelligent generations appearing more slowly and at greater cost in the beginning, but faster and faster every time. This revolution would finally generate an artificial superintelligence explosion, with surprising and unpredictable repercussions in our economies and societies.

This possibility would entail a new technological landscape composed of hundreds of independent AIs with varied abilities and purposes that would evolve and learn through data, algorithms and sensors. They could even create other AIs or improve themselves.

This scenario is what Raymond Kurzweil – futurist, scientist and inventor – named the “technological singularity”.¹ Kurzweil predicted that this would happen between 2029 and 2045, taking into account the acceleration of the exponential technologies, mainly advances in computing processors.

Nevertheless, the term “singularity” was used before in a non-physics context by John von Neumann and Vernor Vinge. Vinge applied this idea to the unexpected irruption of advanced AI with unknown and unprecedented consequences for humanity.

In the physics community, singularity refers to a very infrequent point in the universe, usually inside a black hole’s centre, where physical rules are breaking down as a result of its infinite mass and gravity force.

Everything that exists in the vicinity of a black hole is attracted and trapped towards the singularity once it crosses the virtual line, a no-return frontier, known as the “events horizon”. Nothing, not even light, can escape from the singularity attraction once it crosses this decisive boundary.

Due to this total darkness, a “singularity” cannot be observed directly. In fact, physicists do not really know what happens inside one. In the same way, futurists and engineers cannot see beyond the technological singularity. We can only suspect and guess about the post-singularity effects, and its impact on global markets and economies.

This is the main meaning for the word “singularity” and its stronger implication on our reality. From a physical, technological or financial perspective, nobody really knows what will happen when the singularity is reached. From that moment in time, the most profitable investments will be made by investors with sufficient advanced technology, data sources and skills at their disposal. As a result, investors without AI tools will lose out, because they cannot compete. A new model of investment competencies and skills will be needed to survive in this new ultra-competitive scenario, highly driven by AI.

On the positive side, this kind of technology will be progressively democratized, even demonetized, and largely used by individual, non-professional investors. A new generation of financial services will be provided by banks and FinTech companies: evolved robo-advisors, personal investment assistants, open-predictions services and platforms, insurance for investors and even crowd-investing fully managed by AIs, many of them offered by technological companies and not by banks.

AI technologies are already being applied within banks and FinTech companies. An increasing number of AI solutions – with a very specific purpose – are emerging to resolve frequent tasks and challenges. Most of them are based on reinforcement learning and particularly in neural networks, probably the most extended approach in the investment use cases today. They are achieving encouraging results in fraud and risk detection, predictions, optimization, investment advice and services personalization and – my favourite – anomalies detection.

In parallel, low-latency/high-frequency trading software is being widely used by professional investors and traders, while commercial robo-advisors and financial bots are already democratizing access to intelligent financial advice. All these AIs usually have at their disposal many valuable datasets, thanks to the rise of the API economy and the expansion of open APIs and open data.

If we take into account the latest advances in cloud services, processing and quantum computing, AIs will soon be able to predict almost in real time, making investment decisions faster and more accurate than any human helped by traditional software tools.

How will a “superintelligence explosion” leveraged by this type of AI, specialized in investments, impact the financial industry? The financial singularity could be the answer.

When AIs Rule the Markets: The Financial Singularity

This concept of “financial singularity” describes a future state where, as a result of the convergence of the most advanced computer technologies – such as big data, machine learning, deep learning, blockchains, ultra-powerful processors, quantum computing (now beginning) and cloud computing – we will be able to predict, with minimal errors, all market behaviours and trends. A decisive point from which investors (humans or machines) could determine prices perfectly for every asset in the world, minimizing their risks and increasing the ratio of their correct decisions. It refers to a time when markets could work almost automatically, governed by technology and, consequently, the chance of achieving outperformance (alpha) will go progressively to zero.²

The financial singularity is described by some economists as a theoretical moment in the future when most investment and economic decisions in the world (or even all of them) will be made by superintelligent machines rather than humans, financial organizations or investment experts. If human fallibility is removed from the markets, and asset prices reflect real value, then a utopian period of super-efficient markets will become a reality as a result.

In their book The Incredible Shrinking Alpha,³ Larry E. Swedroe and Andrew L. Berkin pose a decisive question about this environment. Alpha is defined as the outperformance of an investment against a market index: will the alpha opportunity eventually go to zero for every investment strategy in the future?

If not, my question is: could the financial singularity have an equivalent “events horizon”, a non-return line defined by minimum possible alphas?

In order to reach the “financial events horizon” – the point with the lowest alphas ever generated by the financial industry and, therefore, the nearest moment to a hypothetical financial singularity in the financial history – we’ll have to admit that technology will be required to overcome a critical challenge: AIs are mostly reactive, not proactive agents, and they basically take decisions driven by external events detected or data changes. Could they be transformed into proactive intelligent agents, like human investors?

To create a truly proactive AI, engineers need to build “strong artificial intelligence” that exhibits behaviours and reactions at least as skilful and flexible as human brains. This is essentially the ultimate step before reaching the technological singularity. So, both singularities are narrowly connected to each other.

AI in Wealth Management

In my opinion, if the financial singularity happens, it will not occur suddenly. Before the entire replacement of human investors by machines, we will experience a coexistence transition, where AIs will make human investors’ decisions progressively more and more efficient. AIs will learn from human behaviours using massive data: their mistakes, hits, reactions, decisions and results, merging their data with data from others. Investments will be progressively more competitive and they’ll get better returns. Machines will turn into more intelligent agents, training their algorithms and learning continuously from investors, users and markets until definitely becoming capable of substituting us in making super-efficient decisions.

After the coexistence, a second phase – the stage of replacement – might commence, with a progressive and massive substitution, probably not fully, of humans by autonomous AI systems. This period could culminate when machines are able to make their own decisions – even legally and criminally. The AI systems would also show irresponsibilities, bad practices, faults or crimes in the same way than human brokers do.

The Rise of the Financial Events Horizon

I think it will be difficult for us to reach such a state of “financial singularity”. Even if a superintelligence explosion happens in the future, it would not imply the emergence of super-efficient markets managed by hyperconnected machines.

The alphas probably never will be equal to zero, because external tensions – forces generated by the markets, which are chaotic, multicultural and with different interests – will be equalized to internal tensions. These internal forces will be generated by all the intelligent machine feedback, one to another, day by day: the decisions taken by one machine will be the input data for another machine, and vice versa. Machines will be continuously learning about the real world and market behaviours, including decisions and actions from other machines, using all the available data and insights coming from multiple sources.

Progressively, the previous chaos in investments will almost disappear but not completely, because not all the possibilities can be analysed and processed by the AIs, even with tons of data and the best algorithms and processors. They will always need more and more data to improve their predictions: the more intelligent competitive machines are, the more intelligent other machines need to be to compete in this extreme scenario. They will finally accomplish a breakeven, a point where most financial decisions will be perfect to a very high degree, but never 100% perfect.

AIs will pursue financial perfection in every decision (selling at the highest price, buying at the lowest price), but this is mathematically impossible in such an extremely complex scenario. And AI is basically mathematics and statistics. We won’t ever finally reach this utopian and perfect order, designed and governed by super AIs.

It will be interesting to observe if and when we reach financial singularity, with successful predictions and investment advice offered by autonomous agents and intelligent machines. Maybe soon all of us will believe that reaching the singularity is possible.