Humans have a fear of the unknown. When going to a new place, one may feel tense, and in a crisis situation, there may be anxiety about whether "I can get through this crisis well." There is also anxiety about technology. When new technologies emerge, there can be worries and fears about how those technologies will change personal lives and work. Taking artificial intelligence (AI) as an example, since the emergence of AI, many people's jobs have been replaced by AI, and there have been numerous negative opinions about copyright infringement. Is AI really having a more negative impact on people than a positive one, as the public perceives?
▷ There have been significant advancements in the AI industry over the past year. In particular, there have been major developments related to the development and deployment of generative AI models. Generative AI models have transitioned from the experimental stage to being practically applied in various sectors. Key application cases include the following: First, companies like Microsoft (MS) and Google have integrated generative AI into their products to enhance the functionality of search engines and office software. MS's 'Bing Chat' and Google's 'Bard' (now Gemini) are software that have introduced AI-based features to improve user experience. Second, the release of open-source AI models. The release of open-source models like Meta's 'LLaMA' has democratized access to advanced AI technology. This has allowed for broader experimentation and application development related to AI. Third, there is a growing utilization of AI in national security. Recently, Meta has permitted the use of 'LLaMA,' which was previously only available as open-source, to U.S. government agencies and defense contractors. Fourth, the use of AI in drug development is gradually increasing. The biotech company 'Antiverse' is collaborating with global companies to develop antibodies designed by AI. The goal is to make the drug development process faster and more economical using AI. Finally, there has been a change with the introduction of AI in search engines. OpenAI has launched 'SearchGPT,' an AI-based search engine. This signifies a direct challenge to existing search platforms and highlights the competitive landscape in AI-based information retrieval.
- What prompted you to co-author the book 'Thrive: Maximizing Well-Being in the Age of AI'?
▷ The emergence of all new general-purpose technologies (GPTs), including AI, comes with challenges and opportunities. General-purpose technologies redefine industries, render certain companies or jobs obsolete, and generally increase productivity. The difference with AI compared to previous general-purpose technologies is that AI is much more complex and intangible. Because of this, discussions about AI vary depending on who is speaking. There are dystopian narratives suggesting that AI will take over a significant portion of white-collar jobs, as well as utopian stories. Professor Bafna and I wrote the book to inform people about how AI is already embedded in daily life. AI is already widely used in people's health, education, work, and relationships. AI will persist, and for it to operate for social benefit, citizens need to be informed and educated about AI. We published the book to inform people about the 'current state' of AI. For reference, the AI discussed in the book refers to weak AI, which means AI that performs tasks that are difficult for humans due to a lack of expertise or resources, based on machine learning. It does not refer to strong AI, which can think autonomously like humans.
- Specifically, how is AI currently being used in people's daily lives?
▷ People may not realize it, but AI is keeping us safer than we can imagine. For example, AI can detect fake reviews posted on platforms or identify and block unwanted explicit images on dating sites.
- Until now, negative opinions about AI have outweighed positive ones. What do you think is the reason for this?
▷ Human fear of AI arises from various interconnected factors. The evolutionary bias that people have makes them focus more on potential threats than on opportunities when something happens. Because of this, people worry more about the unknown characteristics related to AI. There is no doubt that AI will create new jobs. However, for people, the immediate feeling of AI's replacement of human jobs resonates more than the abstract future opportunities.
Creating AI models is like building a house; it takes 60-70% of the time for foundational work.
▷ I would like to mention the case of the MaMMa Klinika, a breast cancer clinic in Budapest, Hungary. At MaMMa Klinika, two radiologists' breast cancer diagnoses are reviewed by AI. In most cases, AI agrees with the doctors' diagnoses, but occasionally it points out areas that the doctors may have missed for re-evaluation. According to a report by The New York Times, since 2021, AI has discovered 22 cases of cancer that radiologists missed across five centers of MaMMa Klinika. AI continues to be used to save lives.
- What is the basis for AI optimism?
▷ When people first meaningfully encounter AI technology in their work or daily lives, they discover the practical benefits of AI, which forms genuine optimism. In the past, people could experience that they could handle repetitive and tedious tasks that consumed their personal time using AI tools, leading to a positive view of AI. Furthermore, scientific and research applications related to AI reinforce AI optimism. Scientists and researchers using AI systems directly experience and witness that AI processes and analyzes vast amounts of data much faster than humans. There have been groundbreaking advancements in various fields, from drug development to climate science, using AI. This has increased people's expectations that AI can solve significant problems facing humanity. For many early AI optimists, the most attractive aspect was that AI broke down barriers to access various resources and democratized access to them. By removing language barriers or enabling high-quality education regardless of location, AI has been perceived as a technology that brings positive change to human society.
- How can people have a more objective view regarding AI?
▷ Education plays a crucial role in developing a more balanced perspective on the impact of AI. In particular, there should be an emphasis on 'AI literacy' that helps people understand what AI can and cannot do, as well as basic AI functionalities. It is also important to highlight specific success stories where AI has been used to detect diseases early, accelerate drug development, improve access to education, and address climate change issues. The most important aspect of helping people develop an objective view regarding AI is to ensure the formation of comprehensive AI governance that includes ethicists, social scientists, and communities affected by AI. This will serve as a foundation for a more balanced approach to AI development.
▷ AI encompasses a wide range of technologies and applications. It is a synthesis of various technologies, including prediction, explanation, causal reasoning, natural language processing, deep learning, and generative AI. The 'House of AI' emphasizes the need for a solid data engineering foundation to build AI models. 60-70% of the time spent on AI projects is devoted to data cleansing, integration, and curation. The House of AI is a framework that organizations need to understand and apply AI. Just as a house is divided into floors, this framework is also divided into stages, or layers. Organizations should build their 'AI house' step by step, starting from the first floor. They must recognize that without a solid foundation, attempts to apply AI solutions are likely to fail. The foundation of the House of AI consists of three key factors: data infrastructure and governance, computing infrastructure, and AI talent and technology. These elements form the basis of all AI initiatives. As explained in the book, data cleansing, integration, and curation are included in this first-stage process. The data generated here leads to four types of data analysis. In the book, these are referred to as the four pillars. The first pillar is the 'technical pillar.' It involves analyzing hyperdimensional data beyond the three-dimensional limits of human thought to find patterns. The second pillar is the 'predictive pillar.' It predicts what might happen in specific situations based on data. The third pillar is the 'causal reasoning pillar.' It analyzes whether 'X caused Y to happen.' The final pillar is the 'prescriptive pillar.' It finds solutions to specific problems through data.
The next layer involves applying the AI models created based on the analyzed data to real business environments. This is done through processes like deep learning and reinforcement learning based on the data collected in the first stage.
The final layer focuses on AI strategy and ethics. It involves considering what social issues to address and how organizations apply AI. Responsible use of AI is also included in this stage.
- What value does the House of AI provide to companies?
▷ This framework provides a clear process for implementing and applying AI. Therefore, it is particularly useful for companies. The House of AI emphasizes the importance of laying a solid foundation before applying AI technology. In particular, it considers ethical aspects not as something to be addressed after applying AI technology, but as a key part of the AI development process. Based on this concrete process, organizations can better understand what is needed to successfully apply AI and what the process for successful implementation is. At the same time, they can focus on developing and deploying AI responsibly.
- Despite the positive impacts of AI, data security issues remain. What are some ways to ensure the safe use of personal data in AI?
▷ To make the use of personal data in AI-based systems safer, organizations and policymakers should implement a multi-layered security approach that goes beyond basic security measures. The basic steps include strengthening data encryption systems while introducing federated learning (a method of AI learning that creates a final model by exchanging models learned based on their own data without directly sharing data stored in various devices and institutions). Additionally, organizations should establish strong data governance. Clear policies regarding data collection, storage, processing, and deletion should be established. It is also important to set principles of 'privacy-centered design' at the early stages of AI system development. Furthermore, organizations should regularly invest in employee training related to data protection and communicate transparently with users about how their personal data is used. International cooperation is also necessary for the development and implementation of standardized data protection regulations related to AI applications. There should be clear guidelines for cross-border data transfers, a responsibility structure for AI decision-making, and the establishment of security levels by industry. Additionally, companies should invest in high-performance threat detection systems that can identify potential data breaches in real-time and develop plans for how to respond in the event of a security incident.
▷ The most urgent issue for global business leaders is finding a balance between the rapid adoption of AI to maintain (their) competitiveness and applying AI in a responsible and safe manner to protect their organizations from risks. Business leaders are particularly concerned about the need to comply with new AI regulations in different regions, how reliable AI decision-making is when making important business operational decisions, and the significant investment required for AI despite uncertain returns.
These concerns can be alleviated in several ways. First, as mentioned earlier, organizations can establish clear policies regarding AI development and deployment, conduct regular risk assessments related to AI, create transparent documentation for AI systems, and implement a strong AI governance framework that includes ethical guidelines for AI use.
Second, companies can alleviate concerns related to AI by investing in educating their employees about the 'capabilities' and limitations of AI, establishing strong cybersecurity measures, and conducting regular audits of AI systems.
Finally, organizations can start with small-scale AI projects to gain experience and achieve returns on investment before scaling up.
- How do you expect AI to evolve in the future?
▷ AI is likely to evolve to handle complex tasks that require greater contextual awareness and understanding of subtle human intentions and goals. Just as humans solve new problems based on various experiences, AI may also integrate various types of knowledge and capabilities. In specific industrial sectors, AI is expected to advance significantly in drug development and materials engineering. AI will be able to autonomously design and experiment with new compounds. Additionally, in the healthcare sector, AI may evolve to analyze patients' medical histories, genetic data, and lifestyles in real-time to develop personalized treatment plans.
Professor Anindya Ghosh holds a bachelor's degree in Electronics and Instrumentation Engineering from the Indian Institute of Technology (IIT) REC, an MBA from the Indian Institute of Management (IIM) Kolkata, and a Ph.D. in Management Information Systems from Carnegie Mellon University. He has been a professor at NYU Stern School of Business since 2004. In 2014, he was named one of the '40 Under 40 Professors' by the U.S. MBA magazine 'Poets & Quants.' In 2017, he was selected as one of the 30 management thinkers to lead future organizational management by the management site 'Thinkers50.'
[Researcher Yoon Sun-young]
