By Shruti Kharbanda
AI, as we already know, is a remarkable technology. It can help improve efficiency and productivity, generate new content, including text, images, and audio, from existing data. Moreover, AI can analyze vast amounts of information to discover insights that are impossible for humans to detect. Its applications can extend to various sectors, from healthcare and finance to entertainment and education, revolutionizing the way we live and work.
However, despite all its powerful tools, the applications of AI and their impact are increasingly emerging today.
Imagine living in a world:
- Where deepfake content is indistinguishable from reality
- Synthetic identities can orchestrate malicious and disinformation campaigns
- Where targeted scams are crafted with unparalleled precision
- With the increasing prevalence of personalized alibis
- Surrounded by a subtle perpetuation of prejudices for the blatant reinforcement of stereotypes
- Where harmful or radicalizing content is produced by algorithms
As AI matures, the boundaries between the virtual and real worlds are blurring.
Professor Emilio Ferrara has published a 2D model illustrating the harmful applications of GenAI.
Here are some examples of the dual nature of AI:
| Scenario | Opportunity | Danger – risk of misuse |
| Manufacture of historical objects | used to recreate or “restore” historical artifacts or paintings | to create fake historical artifacts |
| Personalized Content Generation | curate content tailored to individual preferences, improving user experience across platforms | create hyper-targeted propaganda campaigns, manipulating individuals’ beliefs or behavior patterns |
| Speech synthesis and cloning | recreate the voices of historical figures, enabling unique educational or therapeutic experiences | fake audio recordings, leading to scams, misinformation and potential security breaches |
| Medical Image Generation | generate medical images for training and education purposes, providing medical students with diverse cases without compromising patient confidentiality | fabricate medical images, which can lead to scams, misinformation, and even potential security breaches, misdiagnoses, fraudulent research, or insurance scams |
| Virtual Reality and Augmented Reality Improvements | enhance virtual reality and augmented reality experiences, making them more immersive and realistic for education, training or entertainment | manipulated realities that distort historical events, spread false information or even create harmful psychological experiences |
| Automated content for social media | for businesses on social media, ensuring consistent engagement and quick responses to user queries | operating multiple social media accounts, thereby creating the illusion of artificially amplifying certain narratives or movements |
| Advertisements | helping businesses create engaging ads and detailed product descriptions | generate misleading advertisements that exaggerate product capabilities or make false claims |
| Financial reports | help financial analysts generate reports, providing insights into market trends and forecasts | generate false financial data or reports, mislead investors or manipulate stock prices |
AI involves complex algorithms and efficient models that learn from large data sets, understand the underlying structures and mimic them in unique ways by generating complex data patterns.
One of the most pressing issues with AI systems is the perpetuation of unfair biases based on flaws in training data or algorithms. Models can learn and amplify existing societal biases around race, gender, age, ethnicity, and more from biased data. This leads to discriminatory and harmful outcomes that can have serious impacts on individuals and groups. Companies can adopt a variety of risk mitigation strategies, such as continuously monitoring AI results, establishing ethical guidelines for AI implementation, and implementing AI-related policies and practices., Promote transparency in AI-driven processes and raise awareness among stakeholders about the potential pitfalls of AI, to ensure informed decision-making. Implement proactive risk-benefit analysis by assessing potential benefits against possible harms, taking into account short and long term implications and preserving our societal values and norms.
Some techniques that can be adopted are:
- Proof of identity:Proof of identity
- Use Aadhar based identity verification
- Multi-factor authentication, biometric verification or digital certificates
- Authentication protocols:identity of an individual, system or entity
- blockchain based authentication
- Token-based systems
- cryptographic methods
- Prominent Public Disclaimer: inform the public about the nature of the content they consume
- For AI-generated content, inform the audience that they are watching, reading, or listening to content generated by an algorithm
- promotes transparency and allows consumers to critically evaluate content
- Content labeling
- visual tags, metadata or even auditory cues to indicate AI-generated content
- Verification of source and provenance: confirm the authenticity and origin of information
- In GenAI, ensuring the provenance of data or content helps maintain its integrity and reliability. Blockchain technology, for example, can be used to trace the provenance of AI-generated content
- Digital watermarking
- integration of a digital signal or pattern into data, allowing their authenticity to be verified or falsification to be detected.
- For AI-generated content, watermarking can help identify and distinguish it from human-generated content.
- It provides a layer of security and traceability, ensuring that any modification of the original content can be detected
With AI, self-regulation is essential. I believe that autonomous, self-managed AI-powered businesses, regardless of the sector, will be based on two underlying principles:
- Human intervention is limited to subjective decision-making.
- The scope of subjective decisions will evolve over time based on the ever-changing business environment, i.e. a continuous improvement cycle of moving decision use cases to AI and adding new decision use cases based on changing business needs.
AI’s ability to augment human capabilities and transform the workforce holds great promise. By replacing low-skilled and low-skilled jobs and creating specialized positions, AI can contribute to a more skilled and dynamic workforce. With careful planning and a commitment to ethical practices, AI integration can lead to a future where humans and machines work together in harmony, driving progress and improving the quality of life for all.
(The author is Shruti Kharbanda – Chairperson, Robotics & AI (Northern Region) – Indian Chamber of Commerce, and the views expressed in this article are her own)
