[New Wave] Why Does AI Seek Blockchain?

AI Crypto, AI Matrix, AICOIN, BotChain, DeepBrain Chain, Neureal.net, Singularity Net, Synapse AI, and others are all companies undertaking projects that integrate artificial intelligence (AI) with blockchain technology. Why are they doing this?

Clearly, AI is the biggest opportunity for the future and one of the fields we must lead in the era of the Fourth Industrial Revolution. However, the side effects caused by AI are also significant, and among them, the most serious is the emergence of 'AI superpowers and their market monopolies.'

The core of AI competitiveness lies in data. Data is the most fundamental infrastructure for training AI, and the more high-quality data is accumulated, the more advanced AI technology becomes. For example, Google, which demonstrated its prowess with the Go AI 'AlphaGo,' owns the world's largest search engine and the video platform YouTube. Google's search engine processes 3.8 million searches per minute, and on YouTube, 300 hours of video are uploaded every minute, illustrating the enormous amount of data Google possesses.

The problem is that companies (or countries) that gain an advantage in the AI field through such abundant data can then collect more high-quality data through AI again, and over time, their market dominance becomes stronger and more entrenched. For instance, AI speakers that perform functions based on users' voice commands send users' voice data to the manufacturer's servers, and the manufacturer analyzes the collected voice data to train the AI on various users' speech patterns and pronunciations. The more data the AI speaker learns, the better it understands user commands, and AI speakers with superior voice recognition sell even more. Moreover, since these companies do not open or share the collected data, the monopoly accelerates, and dependence on them deepens.

What is even more concerning is that companies possessing such AI superpowers could manipulate the system. Researchers have revealed that AI-driven algorithms can have biases. Facial recognition AI software failed to identify women of color, and there was bias against Black Americans in distinguishing criminals. Apple's credit card, Apple Card, faced allegations that its AI algorithm used to determine credit limits discriminated against women. Although it has not yet been proven whether these companies intentionally manipulated the system, if a company monopolizing AI power emerges in the near future, it could manipulate training data to increase AI bias and distort facts.

Blockchain technology can be usefully applied to solve these very problems. The high openness and accessibility of decentralized blockchain prevent specific companies from monopolizing data, and cryptocurrencies provided as incentives to users enable the collection of high-quality data, thereby preventing the side effects of a few large platform companies monopolizing the global AI market. Additionally, the immutability and transparency of blockchain can offer solutions to AI training data bias issues and further contribute to creating 'Explainable AI' by enabling traceability of training data.

Recently, our government announced through the 'AI National Strategy' that it aims to become the country that best utilizes artificial intelligence by 2030, targeting an economic effect of up to 455 trillion won. As mentioned at the beginning, securing AI competitiveness is extremely important at the national level. However, it is also a time when wisdom is needed to carefully monitor whether there are any side effects accompanying it.

Seungjoo Kim, Professor, Graduate School of Information Security, Korea University