I have a piece in today's Financial Express with Parker Karia on the impact of AI (artificial intelligence) on securities markets. The full piece is as follows:
Fast on the heels of development of AI, which has
already resulted in transformations across various industries and professions,
the European Union is the first jurisdiction to introduce legislation that regulates
AI. Broadly, those AIs which pose an unacceptable risk would be prohibited,
whereas minimal risk AI systems are unregulated. The former category would
include AIs that affect people’s rights, biometric categorisation systems,
scraping of facial images from sources to create a facial recognition database,
social scoring, predictive policing, or such AI that manipulates human
behaviour or exploits people’s vulnerabilities. Additionally, the EU’s AI Act
establishes the EU AI Authority, which will be the nodal agency for
implementation and enforcement of the AI Act, which, like the GDPR, has an
element of extraterritorial jurisdiction.
What is actually regulated is AIs that fall into
neither unacceptable risk, nor the minimal risk category. These are high-risk
and limited risk AI systems. High-risk AI systems pose a threat as its uses
include deployment in critical infrastructure, education, essential services,
law enforcement, dispensation of justice, governance, etc. Such systems would inter
alia be required to register with the relevant EU database, assess and
mitigate risks, ensure transparency and accuracy, and more importantly, ensure
human oversight. Moreover, people would have a right to submit complaints about
AI systems and would be entitled to explanations about decisions in relation to
high-risk AI systems that affects the right of the aggrieved person.
Non-compliances with the AI Act will attract fines ranging from €7.5 million or
1.5 percent of revenue, up to €35 million or 7 percent of global revenue,
depending on the violation and the entity.
GPAI
General-purpose AI systems, such as ChatGPT, Gemini,
etc., would face new transparency requirements. Such software, including those
that create manipulated images such as “deepfakes” would have to make clear
that what people were seeing was generated by AI.
AI and the Securities Market
The securities market will not be, or rather, is not
an exception to the infiltration of AI. Although recent breakthroughs have been
largely in the space of generative AI, gigantic steps are being taken to equip
AI with further abilities, as well as expand the data the AI has access to
To put the timeline and development in perspective, till
three decades ago, activities related to trading in the securities market, such
as research, placement of orders, etc., hardly incorporated technology. The
focus on use of technology began with some seriousness upon the introduction of
dematerialized shares. From then, to today, India has taken the lead in
introducing a ‘T + 0’ settlement cycle. Today, any step or activity relating to
trading in the securities market, cannot be imagined without the use of
technology. Now, with the introduction of AI, the securities market is set to
witness another transformation. Today, AI can analyze vast amounts of data,
identify patterns, and make informed decisions in real-time, execute orders at
lightning speed, create investment strategies by itself, etc., as a result of
which, the manner in which trading is conducted, risk is managed, and
investments are made, are rapidly evolving. One of the pertinent concerns in
this regard would be around data privacy.
As AI and AI-generated algorithms permeate various
sectors, including the securities market, regulators face the challenge of
crafting laws that govern these technologies effectively. In the following
paragraphs, AI’s impact on the securities market and its regulation, are
discussed.
Algo Trading & Robo Advisory
Currently, algo trading is defined as trading carried
out through automated means, i.e., order are placed without any human
intervention. Recent suggestions by SEBI to regulate algo trading have received
mixed reviews, while majority views agree with the need for regulating algo
trading, SEBI’s approach towards meeting that end has been the subject of some criticism.
It would also be required to be borne in mind that AI can effectively write
codes based on instructions fed to it. Thus, creating an algo, in the near
future, may not be the exclusive domain of a trained professional in the field
of IT systems. In a situation where the deployment of an AI created algo results
in a violation of securities laws, the question that arises out of this is on
the extent of culpability to be fastened on the person who used AI to create the
algo. The principle that the developer of an AI, or the human behind the
‘machine’ is responsible, exists, however, with the advancements in AI, which
are unpredictable at this stage, the above principle may have to be revisited. As
AI changes the landscape around us, our laws must keep pace to ensure that
rights and obligations of the concerned parties are laid down in advance.
Additionally, robo advisory is presumably going to
take the center-stage in the distant future. For instance, with the vast
amounts of data points analysed in a few moments and investment strategies
being created in seconds instead of weeks, it is not too far-fetched to presume
a considerable shift in the manner in which investment advisory service is
carried out today, which may warrant a revisit to the extant regulatory
framework requiring both strengthening and rationalisation of regulations.
Strengthening for the reasons described above, and rationalising because some
of the restrictions may no longer be relevant with AI generated work product.
Grievance Redressal and Enforcement
AI may eventually be used for dispensation of justice
by appropriately (and safely) integrating it in our judicial systems. In fact,
it was recently suggested that AI may be used to resolve minor traffic
challans, to begin with, upon adequately building up such capability.
Similarly, the securities regulator may consider initiating the process to
develop AI that can effectively monitor, supervise and assist in the
enforcement of securities laws.
Additionally, with the recent focus on online
alternative dispute resolution mechanisms for resolution of disputes in the
securities market, minor issues, depending on the complexity, quantum of money/
assets and the nature of the dispute, AI can serve as an arbiter or mediator.
Pattern recognition and Predictive Analysis
In a potential game-changer for regulators, developing
AI models are becoming increasingly efficient at recognizing patterns, and thus
predicting the ‘future’, depending on the data points that the AI has access
to, and how it is ‘coded’ to ‘think’. While algos are already deployed by
financial sector regulators around the world to identify and/or track
suspicious activity, AI can be of immeasurable assistance in this regard. For
instance, SEBI has in the recent past issued circulars introducing the use of blockchain
to verify information, and to ensure transparency amongst intermediaries and
entities. The integration of AI in such systems can lead to predicting any
defaults, or preventing violations, thus safeguarding investor interest. However,
any such technology should be used with caution, and strict safeguards should
be built around such systems to prevent any misuse.
No comments:
Post a Comment