Rethinking blockchain for business in the era of GenAI

Illustration: TBS

After modern transformer architecture was introduced earlier in the decade, AI technology advanced rapidly, leading to the widespread use of generative AI (GenAI) and large language models (LLMs) in many areas of business and society. Such advancements have enhanced the opportunities for innovation and improvement of the efficiency of business operations.

Businesses can now create their content with quicker turnaround time without compromising on creativity. They can also transform their customer service to an automated one, which resembles human-like conversations with the right degree of emotions through text, voice and visual modes of communication.

Businesses can derive the benefits of GenAI tools which are available in the market. However, the LLMs used by these tools require training data and feedback before they can be used by businesses or individuals. Most of these tools are pre-trained, however as businesses and individuals use them, these tools continue to train themselves through feedback and additional data.

This training and feedback loop can become a significant deterrent for the effective use of the technology if the training data is not authentic. As a result, the trust in the output may get eroded over time and the businesses may lose their brand value.

In the 27th Annual Global CEO Survey conducted by PwC, nearly half (47%) of the Bangladesh CEOs stated that the risk of misinformation is going to increase in their business due to the proliferation of GenAI.

In the same survey, the CEOs had also identified the risks of increasing cyberattacks (57%), legal liabilities and loss of reputation (29%), and bias towards specific groups of customers or employees (31%) due to the proliferation of GenAI in business and society.

Such risks arise due to undesired outputs produced by GenAI tools mis-trained with inconsistent data, what is commonly called 'data hallucination.'

Blockchain technology can address some of the risks arising from data hallucination. Apart from reinforcement learning from human feedback (RLHF) for such tools, the origin of the data should be verified by authenticating the source of data and blockchain technology can play an important role in establishing data provenance while ensuring the democratised sourcing of data.

Democratised sourcing of data can be facilitated by the adoption of Web 3.0– another modern implementation of the internet. Over the last four decades, the internet has evolved significantly from read-only content supply to read-write content exchange, to read-write-own content exchange.

Web 3.0 architecture has enabled the ownership of data to be with the participating population who generate it. Blockchain technology based Web 3.0 architecture could help such GenAI tools to get trained using authentic data from a vast range of sources.

Blockchain technology would be able to provide the much needed intervention layer of data verification and data governance to ensure that GenAI is working with the intended set of inputs. This is how businesses can reap benefits from adopting GenAI while managing the risks associated with the technology's adoption.

Data provenance and data governance mechanisms need to be established for each organisation so that the training data can be tracked and audited before the data gets confirmed. Again, blockchain technology could be an effective enabler for this.

As organisations grow through transformation of their businesses and operations, converging GenAI's adoption with the strength of blockchain technology and Web 3.0 architecture will help businesses in managing risks and sustain the future growth of their operations.

Arijit Chakraborti. Sketch: TBS

Arijit Chakraborti is a partner with PwC.

Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions and views of The Business Standard