In recent months, the public has been captivated by generative AI tools such as ChatGPT, raising concerns about their potential to proliferate misinformation, undermine copyright protections, or pose existential threats. While these questions are certainly worthy of consideration, it's crucial to recognize that other, less attention-grabbing but more immediately hazardous AI systems are already in operation around us.

The 2023 Global Partnership on Artificial Intelligence (GPAI) Summit assembles professionals and enthusiasts from diverse sectors, including business, government, civil society, academia, and international organisations. The summit is dedicated to fostering collaboration among nations in the field of artificial intelligence.

The Global Partnership on Artificial Intelligence (GPAI) consists of 28 member countries, with the European Union included. India holds a founding membership in the Global Platform for Artificial Intelligence (GPAI), a forum dedicated to steering the proper development and application of artificial intelligence.

The Summit features pivotal meetings for governing organisations, alongside side events that highlight AI advancements and facilitate discussions. Side activities include the Global AI Expo, seminars, industry panel presentations, the Game Changer Awards recognizing startups, and a research symposium.

On the third day of the GPAI summit, the session Safe and Trustworthy AI: Adopting Robustness and Security in AI Systems addressed a wide range of topics related to safe, secure, and trustworthy AI. Its objective was to pinpoint critical dangers, challenges, and opportunities for both industry and governments. The gathering aims to unite key stakeholders, fostering the exchange of perspectives among them.

Keynote speakers included Rohit Chauhan, Executive Vice President of AI at Mastercard, and Ms Debjani Ghosh, President of Nasscom. The session was moderated by Dr. Sanjay Bahl, Director General, Indian Computer Emergency Response Team and speakers included Ms. Ramaveda Shri, Advisor to IIT Bombay Trust Lab and Mr. Thomas Scheider, Director of International Affairs, Swiss Federal Office of Communications.

During the session, the discussion emphasised on building and creating safe and robust AI tools that can be trusted. While the pervasive adoption of AI systems in contemporary society is on the rise, these technologies exhibit fundamental limitations and practical drawbacks that can lead to catastrophic failures. Notably, many of the AI algorithms currently in use struggle to ensure safety, reliability, and robustness in the presence of uncertainties. Moreover, the deployment of generative AI systems introduces a host of new challenges, including issues like hallucinations, information leakage, toxicity, and more.

The Moderator Dr. Sanjay Bahl expressed that, “In the 1980's computers became mainstream and when computers became mainstream cost of arithmetic came down to zero, and as a result all business challenges got reframed as an arithmetic problem, for example, word processing, spreadsheets and databases, and it resulted in massive improvement in productivity. In the 1990's internet became mainstream and cost of arithmetic came down to zero and as a result all business challenges got reframed as communication problem, retail became a communication problem. They have just taken a business problem and solved it using communication, it let to massive improvement in productivity. With AI, the cost of production went down to zero. Hence the impact of AI is going to be bigger than the internet and computers”.

Another panelist, Ms. Debjani Ghosh, thinks it is the right time to remind ourselves about that one key promise of artificial intelligence, the promise to accelerate productivity addressing the risk of global slowdown. That is where AI became a must do for the world.

In order to get that boost of productivity back so that you can get back to our pre-covid growth rate we need absolutely large-scale deployment, across industries, across sectors and there lies one of the biggest security risks of AI, Cyber Crime. It is no longer limited to just the tech industry. Today every company is a digital company and if we are a digital company cyber-crime is unfortunately a reality we have to deal with.

While addressing cyber-crime Ms Ghosh also talked about password related attacks which skyrocketed from 3 billion in 2022 to a staggering 30 billion in 2023. She added, “Globally the cost of cyber-crime is projected to cost 8 trillion dollars. If this was an economy it would be the third largest economy after the US and China and it is growing 15% year after year.”

In conclusion, the GPAI Summit 2023 not only shed light on the pressing issues surrounding AI but also provided a platform for global leaders to collaborate, share perspectives, and pave the way for the responsible and secure integration of AI into our rapidly advancing world.