The promise — and peril — of generative AI

Receive free artificial intelligence updates

The author is the founder of the organization Sievea website about European startups powered by the Financial Times

Even though this book was written 37 years ago, Melvin Kranzberg’s First Law of Technology Resonating in today’s rapidly evolving world of artificial intelligence. “Technology is neither good nor bad, nor neutral,” Kranzberg writes.

As historians explain, whether a technology is considered good or bad depends largely on context and time. The same invention may produce different results in different contexts, and these results may change as the situation develops.

For example, the use of the pesticide DDT was initially welcomed as a means of killing malaria-carrying mosquitoes and increasing agricultural productivity. But after environmental activist Rachel Carson revealed how DDT was destroying ecology and poisoning humans, many countries banned its use. Even so, India continues to use DDT for disease prevention, helping to reduce the number of deaths from malaria each year from 750,000 to 1,500.

Two recent and very different artificial intelligence use cases highlight the duality of our latest general-purpose technologies and the difficulty of distinguishing good from bad.

First, the good. Since 1971, Project Gutenberg volunteers have built a valuable public resource: a vast library of free, accessible digital books. As part of its mission to break down “barriers of ignorance and illiteracy,” the program currently provides more than 60,000 off-copyright books. For years, the project has been keen to convert these e-books into audiobooks to benefit the visually impaired, but the costs have been prohibitive. Now, using artificial intelligence, it has generated 5,000 audiobooks at incredible speed and minimal cost and distributed them on Spotify, Google and Apple.

This audio project, Led by pro bono researchers from MIT and Microsoft, 35,000 hours of audiobook recordings were created in just over two hours using the latest neural text-to-speech technology to mimic the qualities and tone of the human voice. “It’s a little more robot-like than a human. But the goal is to get them out of the way quickly and in the least offensive way,” Mark Hamilton, an MIT researcher who co-led the project, told me.

Researchers have demonstrated an even more surprising application that can read an audiobook in anyone’s voice with just five seconds of sample audio. One day, parents might use the app to “read” nightly books to their children, even when they’re not home. But the potential dangers of voice cloning are clear, and the research team is right to be cautious about releasing the technology. “A lot of people in Ayr like to say it’s all roses. But actually, machine learning is a very, very powerful tool that can be used for good and for evil,” Hamilton said.

An example of AI technology being used for evil occurred this month in the Spanish town of Almendralejo, when a group of boys shocked the community by spreading AI-generated nude photos of 28 local girls on WhatsApp and Telegram. Photos of the girls were copied from their social media accounts and then manipulated using generative artificial intelligence applications. Amid nationwide political outrage, prosecutors are now investigating whether any criminal wrongdoing was committed.

Gema Lorenzo, mother of a 16-year-old son and a 12-year-old daughter, said all parents are worried. “You worry about two things: If you have a son, you worry that he might do something like this; More worried because it’s a violent act,” she told the bbc.

The promise and danger of generative AI is that it is now so readily available and can be deployed at incredible speed and scale. OpenAI’s ChatGPT chatbot has been tried by more than 100 million users within two months of its release. Tech companies are now building more powerful multimodal models that combine text, images, audio and video.

Kranzberg writes: “Many of our technology-related problems arise from the unforeseen consequences of the large-scale use of seemingly benign technologies.” This is true of social media, and so is generative artificial intelligence. in this way. As with other technologies, what will follow will be a chaotic period of behavioral, social and legislative adaptation.

Sequoia Capital Report Articles published this month suggest that generative artificial intelligence is still in its “awkward teenage years.” This industry certainly has a lot of growth to do, and it requires the involvement of parents.

Svlook