top of page
Writer's pictureTheCyberDiplomat LLC

Engineering a Generative Artificial Intelligence that Empowers Democracies

On the International Day of Democracy and Engineers Day, Sanjana Rathi, the Chief Executive Officer of The CyberDiplomat, shares her insights on the intersection of engineering and AI development for the betterment of democracies and humanity. Her comprehensive article delves into the multifaceted role of engineers in shaping the societal landscape and highlights the ethical and foundational elements essential for the advancement of a more promising future.



GEN AI Generated for the following prompt : Gen AI impacting democracies and humanity


The rise of generative artificial intelligence (AI) and the expansion of AI capabilities have raised numerous questions about the impact of technology on body politics and information, as well as trust within a functioning democracy. Even before the 2020 breakthrough in generative AI, many nations worldwide were already grappling with issues related to misinformation and disinformation campaigns, especially during the COVID-19 Pandemic. However, generative AI introduces new possibilities for both intentional misuse and unintentional misinterpretation of information, with far-reaching implications for election cycles, the empowerment of dangerous nonstate actors, the proliferation of misinformation and disinformation, and the potential to undermine electoral processes.


The hallucination of the Truth

“Many lies cannot be seen as truth.”

The use of large language models (LLMs) in various digital platforms and communication channels has the potential to impact democratic processes significantly. LLMs can perpetuate and amplify societal polarization by disseminating false or inaccurate information on a massive scale. This phenomenon can lead to a decline in public trust in media and essential institutions, thereby threatening the integrity of democratic systems. Moreover, there is an assumption in these LLM models that people are rational. At the same time, behavioral analysis of people emphasizes that human behaviors are irrational. For example, based on the theories of behavioral economics, the study of practical cases and data concludes that people are irrational many times in their daily lives. Moreover, LLMs go by the model of respecting the majority. Therefore, even if the truth is rare, LLMs are designed to support false narratives and information, running a risk of corrupting the system and further complicating the process of disseminating quality rather than simplifying it.


Good Data — Bad Data


Poor data quality significantly hinders the effectiveness of AI models. If AI is trained using incomplete, inaccurate, or biased data, it will lead to compromised outcomes. For example, if an AI model is trained on data with vital information missing or containing erroneous values, it will struggle to provide reliable predictions or make sound decisions.


Deception


Deepfakes are synthetic media that have been manipulated to deceptively alter the appearance, voice, and disposition of an individual. Creating deepfakes or AI-generated images or videos is alarmingly simple and inexpensive. However, this technology is highly deceptive and gives rise to serious integrity issues. Additionally, it significantly complicates law enforcement of technology policies and laws. AI-generated images refer to synthetic imagery created through an image generation tool like DALL-E or Midjourney. Even short samples of an individual’s speech can be cloned using voice cloning to generate synthetic speech from text inputs.

While many articles discuss AI’s concerns and its threat to democracy, we at The CyberDiplomat highlight the solution for engineering a better future with responsive AI becoming ‘responsible AI.’ We are listing possible steps that will lead to this.


1. Going Back to Basics


The basic foundation of a society is built on relations, principles, and the quality of people we live with. Our coexistence and connection with nature nurtures creativity and enhance the quality of life. We must understand that significant innovations that succeeded were inspired by nature. For example, the vision of flight was born from seeing a bird fly. Gravity was born from a detailed observation of an apple falling from a tree. For building a great AI system, have we analyzed the mind of a human being that has no prejudice, no bias, no affiliates, and in absolute truth? Still waiting!



2. Focusing on Human Resources

We must build ‘good’ and sensible human beings to strengthen our AI capability and capacity building. By investing in technology more than human resources, we are jeopardizing the possibility of making human beings that would build great technology. Thinking and creating a beautiful LLM model-based AI is a state of the art. However, our society feels that art negates rational thinking and decisive decision-making. This narrow thinking, amplified by the notion that art and humanities don’t hone any skill, leaving a person unemployable, needs to be revised. We need to focus on building a system that teaches the fundamental values that hone human geniuses.

3. Focus on Algorithmic Optimization and Transparency

To promote transparency in algorithms and optimization, allowing the public to observe and review the algorithm by disclosing its internal workings is crucial. This will help eliminate doubts, anxieties, and fears and ultimately enhance trust in the algorithm.


Transparency mechanisms must be implemented neutrally. Researchers have uncovered “dark patterns” of transparency that can create a false sense of security for users and deceive them into believing that the system is trustworthy or fair, even when the underlying model is biased against minority groups.


Two examples of dark patterns include:

1. Providing users with large amounts of irrelevant information makes a model appear more fair or trustworthy.

2. Misrepresentation of a message through data visualizations using techniques like exaggeration or understatement, as research has shown.

4. Knowledge about AI safeguards every human being in a Democracy



Understanding AI safeguards, usage, and challenges is essential for the safety and security of technology implementation. This AI curriculum does not involve teaching hardcore engineering subjects to non-engineers but instead aims to educate individuals about how AI is created and its limitations. The course must encompass AI’s philosophical and social aspects to ensure universal comprehension and implementation. Intelligence is Knowing that Artificial Intelligence (AI) is not Real Intelligence (RI).


Overall, the influence of AI on Democracies and the 2024 elections will be shaped by how it is used, regulated, and managed by various stakeholders, including technology companies, political campaigns, and regulatory bodies. It will not be so much realized in 2024, but it will be when multiple nations go on elections again in 2028. It will be crucial for voters to stay informed and critically evaluate the information they encounter to ensure the integrity of technology, AI, and cybersecurity in the democratic process. So, are you prepared?


The CyberDiplomat is shaping cutting-edge technology and its effects on society. If you feel our service can impact your work or business, please get in touch at sanjana@thecyberdiplomat.com.



 

About the Author : Sanjana Rathi is the CEO & Founder of CyberDiplomat LLC, USA and Cyberdiplomat Tech Pvt. Ltd., India, having 10+ years of experience in the field. The CyberDiplomat is the result of her professional and educational qualifications in multiple disciplines, with her focus being ‘Cyber’. She is a TATA scholar holding M.A degree in Political Science Security & Diplomacy from Tel Aviv University, & MSc.Degree in Management of Information Systems & Digital Innovation (MISDI) from London School of Economics & Political Science(LSE). She is a Computer Science Engineer from VTU, Belgaum, also holding a Diploma in Cyber Law and Forensics from National Law School of India University, Bangalore. She is also a Certified Information System Security Professional.


In the past she has worked with International Law Enforcement, Think Tanks, Private, Public sector enterprises and academia. Having worked in all domains of cyber, she is passionate and committed to the field — which she brings to CyberDiplomat LLC.


Sanjana is also a Drone Pilot and having a Diploma in UAV & Drones from the Indian Institute of Drones.

2 views0 comments

Comments


bottom of page