AI's Promises, Perils, and the Path to Wider Adoption

Key Insight

The rapid advancement of AI in 2023 introduced significant concerns, including the potential for an internet 'polluted by low-quality, AI-generated text,' or 'slop,' with studies warning that training future models on excessive synthetic data could lead to their degradation. The approaching 2024 presidential election intensified fears of 'deepfakes'—highly realistic AI-generated images, audio, and video—while a fake photo of a Pentagon explosion demonstrated AI's immediate capacity to influence U.S. stock markets. Safety concerns also materialized with autonomous vehicles, notably in October when a Cruise driverless taxi dragged a pedestrian 20 feet, leading to a California-mandated halt and subsequent nationwide suspension of its driverless operations.

Generative AI posed a clear threat to several job categories throughout 2023; industries like advertising, formulaic content creation, and customer service faced potential displacement. For instance, Duolingo reportedly replaced contract translators with AI. A major ethical and legal battle emerged over the widespread use of copyrighted intellectual property from writers, artists, and musicians to train generative AI models without compensation or attribution. The New York Times filed a lawsuit against OpenAI and Microsoft, alleging copyright infringement and presenting samples of verbatim plagiarism by chatbots, underscoring the critical need for creators to be compensated for their work.

Despite the perils, AI demonstrated immense potential for societal benefit. It served as an early-warning system for natural disasters like earthquakes, hurricanes, and wildfires, and the IRS utilized it to identify wealthy tax evaders. In healthcare, AI facilitated a paraplegic's ability to walk again in Switzerland via a 'digital bridge,' detected Type 2 diabetes with up to 89 percent accuracy from voice analysis in Canada, accelerated drug research, and consistently outperformed medical experts in reading mammograms and other medical imaging, with Google's Med-PaLM2 model showing superior diagnostic performance. However, companies like Inflection AI faced substantial challenges in user adoption, with their Pi chatbot capturing only 2 percent of the market share. Inflection focused on developing smaller, more efficient AI models and expanding Pi's capabilities beyond conversational AI to include coding, reasoning, and action-taking, aiming to become a comprehensive 'personal AI' for any user need.