Forging Uncharted Horizons: Navigating the Risks of AI

Artificial Intelligence (AI) has the potential to radically alter almost all spheres of human life. This is because it allows for quick processing of vast amounts of data and makes predictions or decisions that would be otherwise rather impossible for humans to do in a reasonable amount of time. AI holds the potential to substantially improve efficiency, accuracy, and safety in many fields beyond human capabilities. But is that enough?

AI also poses several dangers, such as the potential for bias and discrimination if the data used to train AI systems, made by humans themselves, is not diverse or representative of all populations. There is also the risk of job loss as AI takes over tasks previously performed by humans. Additionally, there are widespread concerns about AI being used for malicious purposes, such as cyber attacks or weapon-building and spreading misinformation.

To maximise the benefits of AI while minimising the risks, it is important for stakeholders like business magnates, technologists, policymakers, and society as a whole to consider the ethical implications of AI and ensure that it is developed and used responsibly.

AI and human intelligence: Are they the same?

Artificial intelligence and human intelligence are two distinct types of intelligence, each with its strengths and limitations.  One of the key differences between artificial and human intelligence is the way they learn. AI systems attempt to replicate forms of human intelligence. However, even though AI can process vast amounts of information quickly and accurately, it can still come with its own limitations.

Risks posed by AI:

The fears  that AI can take over all human jobs can be misleading. As Adiraj Chakraborty noted in his 2019 Letter to EPW, ‘Artificial vs Human Intelligence’ “Most of the jobs that we can think of require a multiplicity of skills. From technical expertise to intuitive mastery, jobs generally require a host of cognitive and socio-behavioural skills that are interconnected.”An EPW Engage article titled, ‘Interrogating the AI Hype: A Situated Politics of Machine Learning in Indian Healthcare’ authored by Radhika Radhakrishnan highlights the need to prioritise a more human-centred approach to technology development and deployment, to ensure that the benefits of technological advancements are equitable and accessible. This is due to the possibility that AI can, she warns, exacerbate existing inequalities or perpetuate the profit-driven interests of private corporations at the expense of marginalised communities. She suggests that “instead of asking, “How can AI solve this problem?” it is most worthwhile to ask  “What problems can AI solve?” This has at least two advantages: one, it makes us focus on using AI-based technologies to solve only those problems that are within the scope of technology to solve, and two, it incentivises us to find need-based solutions as opposed to market-driven capitalist solutions to identified problems.”

Others even question the permeation of AI in creative spheres such as that of art. Neerej Dev and Vipula P C in their 2023 letter ‘Artificial Intelligence and Art’  show that there is great scope to philosophically investigate and “question the ethics of using artificial intelligence for commercial purposes without giving proper credit to the original creators.”

Suchana Seth, in her 2017 article, ‘Machine Learning and Artificial Intelligence Interactions with the Right to Privacy’ explains through the example of algorithms are being used “increasingly by state and corporate actors for surveillance that falls well outside the domain of law enforcement.” She shows the very real challenges cast  by artificial intelligence in the realm of human privacy. In a similar vein, Gaurav Pathak, in his 2022 letter titled, ‘Clearview AI: Dangers to Privacy from the Private Sector’, argued, Gaurav Pathak exemplifies this in the context of Clearview AI, which is a United States (US)-based company that downloaded over 10 billion facial images from the internet. He writes, "a citizen can challenge the use of Clearview AI or any other facial recognition technology by the state, but what remedies does one have against the existence of a private database such as the Clearview AI? The Supreme Court in the Pegasus case (Manohar Lal Sharma v Union of India and Others, Writ Petition (Crl) No 314 of 2021, Order dated 27 October 2021) has said “the right to privacy is directly infringed when there is surveillance or spying done on an individual, either by the State or by any external agency.” It even cautioned by saying, “it is undeniable that surveillance and the knowledge that one is under the threat of being spied on can affect the way an individual decides to exercise his or her rights. Such a scenario might result in self-censorship.”

On similar lines, Tripti Bhushan, in her 2023 letter titled ‘Negative Impact of ChatGPT in Research’ wrote about the dangers posed by newly introduced technology of generative AI, such as ChatGPT. She contends “ChatGPT is capable of generating highly convincing responses to questions and prompts, which could make it difficult for users to discern between accurate and inaccurate information.” Further, ChatGPT can disrupt the creation of independent and original scholarship and give more space for plagiarism to arise. Bhushan iterates "The [ChatGPT] model has the ability to generate responses that are highly convincing and could be used to manipulate individuals or spread propaganda. This could have serious implications for democracy and could potentially harm individuals and communities. For example, if a political organisation were to use ChatGPT to generate responses that were designed to influence voters, this could lead to a breach of trust and could potentially harm the democratic process."

Additionally, large language models, such as ChatGPT, could exacerbate the concentration of information in the hands of a few, with the resources to develop and deploy such models. This trend goes against the principle of decentralisation, where power and access to information are distributed across  broader networks.

Conclusion

These examples point towards the fact that there is a need to tread forward with caution. Artificial intelligence needs to be imagined in a way that it does not override ethical norms, it should be more inclusive, and we should aspire to make data and algorithms more accountable, by enabling transparency and initiating safeguards in its creation and deployment.