AI Drivers
The intractable issues posed by AI are currently driven by many large forces and interests. These driving forces for AI are classified against a power index* a number on a scale of 1 to 10 indicating its power to impose global change in the current state of AI development. Commercial interests of Multi-National Corporations (MNCs), Governments, Religions, Political ideologies, Strategic interests and Venture capital. The histogram is represented in fig1 with the Pareto line. Surprisingly, people who are directly affected do not figure in the list of drivers.
Fig 1: Current Driving Forces and global impact
How is AI centered?
AI is highly pervasive, hence its main intent and focus of design becomes critical. This may be called the “centricity” of AI. The spectrum of AI centricity can range from offence, defence, commercial, human capability supplement to fully human-centric. Is the AI center close to innovation (meaning highest emphasis on profitability, high valuations and growth) or is it centered on the support for the ethics, rights and privacy of human beings? The drivers of AI (Fig 1) increase the power and control of the driving agency over their distinctive customers and markets. The important observation here is that with the coming of AI, the locus of control of new AI discoveries has increased exponentially. This means that if a new AI technology is discovered it becomes immediately impactful on a host of industry verticals and personal use cases on a large scale. It is very difficult to list the areas of overall impact, thus the need for regulated and safe AI design practices1 . This should worry us because of its unintended or intended consequences affecting humans.
Governments and AI investments
Just like the Sarbanes Oxley Act (SOX) through which conflicts of interest in auditing and consulting businesses of big consulting companies were regulated, governments and their functioning must be kept separate from private sector interference. It is the government which exists to protect the people not private sector commercial leadership. Else the interests of the people will be compromised by weakening regulation to favour commercial interests as against people interests. The growth of monopolies is also an undesirable consequence.
Governments should invest and build adequate competencies in house for testing AI systems for compliance. They should also build capabilities and skills for meeting their obligations which include pursuing investments and policies to encourage the rapid adoption of AI technologies. Re-skilling and up-skilling of personnel is an essential milestone in the journey of AI adoption.
Any nation that has the best AI capabilities will be in a winning position in the future. Due to this, there is a race for possession of the most advanced AI capability by nations. What can be the outcome of this? The mass rush for developing AI is good and initial differences will be flattened out over time. Nations are building their own LLMs which would suit their own needs. Yet, it should be remembered that it is scientific discoveries and research capabilities that drive innovation in this world. Transformer architecture, neural networks and self- attention mechanisms form the core discoveries that enabled LLM technology to be developed and commercialized. Neural networks and statistical models enable pattern recognition. Statistical methods, differential calculus are core to analytical models. Algorithms like decision trees, regression etc facilitate categorization. Hence, pure Science, physics and mathematics are the womb for the birth of AI innovation. Discovering new learning algorithms, step change processing technologies like quantum computing, the designing capability for chips, data centers, cheap and plentiful power and more constitute some of the scientific capabilities required. The countries with the best scientific capabilities will lead in the long-term highlighting the important need for investments.
Human Beings
Human beings are different- our approaches to solving problems, planning a speech, or reasoning, leading, admonishing or facing any situation is different and unique. Therein lies our uniqueness. This uniqueness is what creates a perfect balance in this chaotic world of ours. AI must not upset this balance. This is one of the main reasons why AI needs to be regulated. In addition, when it comes to personal data the following is non-negotiable:
1) The collection of Core Personal Data (CPD)2 from individual mobile phones and other personal devices for creating Personal AI Models (PAI)2 should be banned by law till the time we are able to develop adequate technology safeguards to monitor CPD and PAI processing.
2) When AI and Data technologies are mature enough, we may permit PAI model training and processing on the personal device itself without the CPD being removed from the device. The operation should be through APIs with real-time monitoring and logging.
Peaceful coexistence, societal harmony and thriving business environment underpin successful AI adoption.
National safety measures
Nations should carefully decide whether to provide access to global AI products and services within their jurisdictions. India should stand firm on its DPDP Act and the EU on their GDPR. Governments have the bounden duty to protect their citizens by preventing access to their markets to AI products and services which are non-compliant. Banning entry is a more effective solution that allowing entry and then fighting legal battles for years on end after the damage has been inflicted.
There is a lack of knowledge of the benefits and harm of AI systems. Educating people on AI for voluntary action and exercising their choice is an important component of safety. The novelty and hype of a new toy in the hands of a child syndrome is being exploited by AI companies to market and spread the use of AI tools. It is not practical to read the fine print of the terms and conditions of AI products. It is the responsibility of Governments to ensure safety of their citizens. This website has been providing human centric, fair and unbiased information and learnings for increasing the AI knowledge of people for the last three years. Ultimately, people should have the freedom to use AI in accordance with their individual preferences. The AI for Good initiatives being conducted by UN is a good starting point, yet more needs to be done to bring nations worldwide to work together.
Conclusion
It’s time to have a series of debates in the United Nations to better understand the global dynamics of AI. The United Nations was formed to prevent war and maintain peace. This is the first time in world history that we have a modern technology, which is literally all encompassing and developing at an uncontrollable growth rate penetrating every aspect of human life. A technology which could assist or divide the nations and people of this world. Discussion should focus on preservation of diversity and freedoms existing in the world and on mechanisms to guide the development and application of AI technologies in a Human Centric 1 manner. This is a new agenda for the United Nations to pursue. After all it is the only global body, we have with the vast capability to bring about a dialogue on global issues confronting the world.
Note: * The power index is an estimation made by the author based on his personal understanding and is time dependent.
Reference:
- Applied Human Centric AI by Rajagopal Tampi.
- Personal AI Models (PAI) – An ethical violation of AI modelling people – aipathfinder.org – Pathfinder for Artificial Intelligence
Disclaimer: The opinions expressed in this article are personal opinions and futuristic thoughts of the authors in their genuine quest for designing AI systems in a Human-Centric and Explainable manner. No comment or opinion expressed in this website is made with any intent to discredit, malign, cause damage, loss to or criticize or in any other way disadvantage any person, company, governments or global and regional agencies.