Review | We can’t stop AI, but here’s how we might model its effects

The tech industry loves trying to convince us that the next thing is the real fix, transforming society and generating billions in the process. Sometimes, as is the case with personal computers, the Internet, smartphones and social media, the hype is justified; other times do you remember the NFTs? the Metaverse? Not so much. It’s easy to get lost in the daily froth of gadgets and gimmicks, booms and busts, winners and losers.

Take ChatGPT, for example. It took the world by storm, setting the record for fastest growing app in history and spawning countless clones. But less than a year after its initial release, cracks are emerging: the cost of running a chatbot has become a serious issue. The trend for chatbots to confidently tell lies doesn’t look set to go away anytime soon. And Microsoft’s plan to reinvent web search with chatbots hasn’t even dented Google’s market dominance.

Are chatbots ushering in a new era of civilization or are they yet another overrated fad? It’s too early to tell. But when we step back a bit and look beyond the daily ebbs and flows, it’s easier to see the broader currents of technological change. Chatbots are just an application of large language models, which themselves represent just one corner of contemporary artificial intelligence. And AI is a big part of a massive wave of technology that we’re just starting to experience.

This Coming Wave is the subject and title of an expansive and thought-provoking new book by Mustafa Suleyman (written with Michael Bhaskar), co-founder of the leading AI lab DeepMind, which was acquired by Google in 2014.

I like to think of AI as data science 3.0. Traditional statistics, from means and medians to pe values ​​to significance tests, have revolutionized science, medicine, and many aspects of government and business operations, particularly from the 19th century onwards. The early 2000s marked the beginning of a second period, even more dependent on computers for processing large amounts of data (big data). High-resolution statistics have become the engine for things like tech giants’ predictions of what consumers are likely to buy next, and political campaigns like Barack Obama’s 2008 campaign team, which decided how to focus its efforts using unprecedented fine-grained voter information.

In the current third wave of data science, the emphasis is shifting from making predictions to automatically acting on them, and from analyzing data to generating it. Any major changes that happen to society in the coming decades are likely to be data-related in some way. And whatever new technique drives these changes will likely be labeled AI, no matter how distant it may be from what we call AI today.

Suleyman does not describe AI as I have here, but he equally sees it as part of a larger technological age, which is an integral part of genetic engineering, especially gene editing and synthetic biology. In these currents there are also other potentially revolutionary technologies, such as quantum computing and fusion energy. Suleyman argues cogently that none of these technologies develop in isolation; they proceed synergistically, as progress in one area stimulates progress in others.

Suleyman sees a striking commonality in the technologies that make up his next wave: They proliferate power, and they do so by reducing the costs of information-driven action. This, he believes, sets it apart from the previous wave of Internet-related technologies that reduced the cost of transmitting information. While the world is too messy to fit neatly into simple summaries of this kind, I find Suleyman’s framing to be reasonable and useful enough: He looks less at the individual technologies within a wave, he suggests, and more at what these technologies allow people to do.

Suleyman argues convincingly that tremendous progress for humanity is possible with what is coming, but he also argues that this wave will shower us with devastation unless we work harder to direct it. Whether it’s the deliberate use of powerful tools or accidental accidents of unprecedented magnitude, there’s a lot that could go wrong.

While fanciful doomsday prophecies are a popular concern in and around some technological circles, this book provides a well-founded analysis. Instead of the familiar Hollywood robot takeover list (HAL and Skynet disappear!), you’ll find balanced discussions that emphasize the sociopolitical and socioeconomic context in which the technology develops and exists.

Suleyman also deviates from the more common tech industry line in the way he impressively draws on the past to help us understand the present and prepare for the future. Historical vignettes of technological progress, from the Industrial Revolution to the combustion engine to the dawn of the Internet, are captivatingly woven throughout the book. As these examples demonstrate, technological waves are nearly unstoppable and we shouldn’t want to stop them anyway, because technological stagnation is not the answer. As he astutely writes, modern civilization issues checks that only continuous technological development can cash.

It is especially impressive and welcome that Suleyman includes a broad and in-depth discussion of concrete, practical steps we can take. His suggestions are extraordinarily broad and balanced. He strongly rejects the hyper-libertarianism of tech moguls like Peter Thiel and advocates strong regulation and international cooperation, but acknowledges the short-sighted nature of modern governments and the myriad ways regulation fails. On economics, he doesn’t go as far as some scathing criticisms of the capitalist underpinnings of artificial intelligence, but he goes much further than most of the tech industry when he discusses the role of financial incentives in encouraging dangerous risk-taking. He also offers some interesting ideas on tax policy and corporate restructuring that deserve more attention.

Suleyman falls into a few traps common to tech leaders, such as assuming exponential progress when it isn’t, underestimating the human cost of building AI systems, and highlighting his efforts to sound the alarm about AI by conspicuously failing to mention many other individuals. they’ve been doing it for years. It is especially shameful that none of the women featured in this recent Rolling Stone article are mentioned, or referenced, in Suleyman’s book. And he takes a questionable stance on open source software, suggesting that AI systems shouldn’t be deployed on a large scale, even though many experts believe this is the best way to find out about their problems so they can try to fix them. But these issues do not detract from the overall value and importance of the book.

It remains to be seen whether ChatGPT will end up central to the next wave or simply as debris washed ashore by the technologies that really matter. Instead of focusing on which apps will stand the test of time and which startups will be successful, we should look up and recognize what is fast approaching and that there are many things we can do to prepare. Suleyman provides much-needed and unusually thoughtful guidance that is expansive, historically grounded, and captivatingly written.

Noah Giansiracusa is professor of mathematics and data science at Bentley University and author of How algorithms create and prevent fake news.

Technology, power and the greatest dilemma of the twenty-first century

By Mustafa Suleyman with Michael Bhaskar

A note to our readers

We participate in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide us with a means to earn commissions by linking to Amazon.com and its affiliate sites.


#Review #stop #heres #model #effects
Image Source : www.washingtonpost.com

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top