The dangerous future of generative AI – the crypto asset community that advocated decentralized AI[Davos Conference]| CoinDesk JAPAN

A look at the list of companies leading the generative AI race illustrates the risk of concentration of power in this technology and how blockchain’s decentralized data management model can mitigate it.

Big tech leading AI development

The five most prominent companies currently getting into AI are some familiar faces. Microsoft, Alphabet, Amazon, Apple, and Meta.

These are the Internet platformers that have dominated Web 2.0 for the past 20 years. These five companies are investing billions of dollars in generative AI, through large investments in startups like OpenAI and Anthropic, as well as their own projects.

All five companies are in the top seven in the ranking of companies by market capitalization. The total market capitalization of the five companies is just under 10 trillion dollars (approximately 1,480 trillion yen, equivalent to 148 yen to the dollar). By the way, NVIDIA, which ranks 6th in market capitalization, has its GPUs actively purchased by five companies for the development of large-scale language models (LLMs) for generative AI. Adding the company’s market capitalization, the six companies account for more than a quarter of the S&P 500’s market capitalization.

It is strange, but understandable, that the only company in the top seven other than these six is ​​Saudi Aramco, Saudi Arabia’s national oil company, which has the third largest market capitalization. Data, which gives Big Tech an edge, is often described as “new oil.”

The five Big Tech companies are in a dominant position in the race to develop generative AI because of the vast amounts of digital data they have about us, the data we use to train LLMs.

The Big Five search engines, social media, browsers, operating systems, and cloud computing services extract vast amounts of data about our online activities and the social relationships they reveal. We were their quarry for finding generative AI.

Current model risks

The business model of surveillance capitalism that pervades the internet economy allows these companies to use their products to create new algorithms to target us, providing constant stimulation for us to benefit their business interests. Secretly instructed him to take action.

Over time, they repeatedly tweaked a set of operational tools to keep advertisers, app developers, and corporate IT departments paying for their services, and to keep us endlessly engaged with their platforms. It began to.

Six years ago, Facebook President Sean Parker inadvertently admitted that this was a deliberate plan aimed at “exploiting vulnerabilities in human psychology.”

The aforementioned market capitalization figures show that this model was well aligned with the interests of the platform’s shareholders. But there is now irrefutable evidence that this model has strayed far from the interests of society as a whole.

With youth suicide increasing by about 50% since 2008, the US Surgeon General warns that exposure to cyberbullying and other harmful behavior threatens young people’s mental health.

On the other hand, when every issue creates a torrent of abuse between competing interest groups, it is difficult to ascertain the facts, and by extension, it is difficult to resolve urgent issues such as climate change or the conflict in Gaza. It has become.

As Frank McCourt and I argue in our recent book, Our Biggest Fight, today’s Internet economy is seriously undermining the health of democracy.

Why bring in such a destructive oligopolistic model in the AI ​​era, where data-driven algorithms have a greater influence on our lives? Why give a centralized corporate owner of an AI infrastructure absolute control over all important information about our true nature as humans?

Of course, platforms will fight tooth and nail to protect their rights to use their data. But we’ve reached a point where we need to recognize the data as our own. It is too dangerous for critical information to be monopolized and secretly manipulated by companies that have already demonstrated the ability to harm us.

The need for decentralization

We will not discuss here how to realize a model in which data and content are managed at the end (user side) rather than at the center of the network.

But changes in data management models are on the horizon anyway. The New York Times’ lawsuit alleging that its articles were captured by Microsoft-backed open AI will prompt many companies that manage digital content to withhold new material from AI companies. is expected.

This would pave the way for AI models running on more decentralized systems, where training data is only used with the owner’s consent.

This requires a decentralized tracking approach, such as that enabled by blockchain. This is to ensure consenting owners that their data and content is used as described, and to ensure that sensitive information is not subject to AI-driven “deepfakes.” We need a verification system where people trust open source protocols that are resistant to censorship, not Big Tech’s promises to “do the right thing.”

The other day, at the World Economic Forum (Davos Conference) held in Davos, there was a discussion among people involved in crypto assets (virtual currency) (most of whom were active outside the security of the Davos Conference). No wonder one of the most talked about topics was the intersection of AI and blockchain.

They include the Hedera Hashgraph-proven data validation system presented by Jonathan Dotan of Starling Lab, and the data validation system of Erik Voorhees and David It was buzzing with new developments, including a distributed computing project called MorpheusAI led by David Johnston.

But the discussion was largely excluded from other AI programs during Davos. At Davos, many global companies pitched their solutions to save humanity from AI. Tata Consultancy Services, for example, has put up a giant sign that says, “The Future is AI. The Future is Humanity.”

This is unfortunate. It is now urgent that the mainstream listens to the blockchain community’s decentralization discussions.

｜Translation and editing: Akiko Yamaguchi, Takayuki Masuda
｜Image: Growtika/Unsplash
｜Original text: At Davos, Crypto Pushes Case for Decentralized AI