He was right about AI. His next predictions are wild.

Summary

The video reviews Kevin Kelly’s 2014–2016 predictions about AI, compares them with developments described in the subtitles, and outlines his newer forecasts.

Earlier Predictions and How They Compare

• AI as a cloud utility: Kelly expected businesses to buy AI capabilities from cloud providers, much like electricity, rather than build the underlying systems themselves. The narrator sees today’s cloud AI offerings, including Amazon Bedrock, as evidence that this prediction was broadly right.
• A few providers controlling AI infrastructure: Kelly anticipated that a small number of large, general-purpose AI services would underpin many startups. The video argues that this could concentrate influence among infrastructure providers, even as many companies compete to build applications.
• AI replacing Google Search as Google’s main product: Kelly predicted that AI would become Google’s central product by 2026. The narrator says this has not happened in revenue terms: Search still brings in substantial income, though the familiar search experience has changed.
• Conversational interfaces: Kelly foresaw people asking cloud systems questions in everyday language and following up for clarification. The narrator considers this prediction accurate, while stressing that users still need to judge whether AI explanations are trustworthy.
• Narrow AI specialists: Kelly expected many systems to be highly specialized. The video says modern foundation models have proved broader, handling many kinds of tasks within one system.
• People working alongside AI: Kelly expected AI to automate tedious work and make human–machine collaboration more economically valuable. The narrator questions whether current developer practices, such as writing extensive instruction files for coding assistants, really constitute a meaningful new skill.

Kelly’s Newer Predictions

• Powerful local AI: Models comparable to today’s leading systems may run on personally owned hardware within a few years. Training a similar model from scratch could become affordable to an individual in roughly a decade. Proposed benefits include greater control over data, costs, and business tools.
• Always-on AI companions: Within about five years, smart glasses or similar devices might provide companions that continuously observe and listen, learning users’ habits and preferences. The narrator highlights the privacy and surveillance risks.
• Software designed for AI agents: Companies may adapt their codebases so AI coding tools can work with them more effectively.
• Agent-to-agent commerce: AI agents may increasingly buy services from other agents, possibly reaching a scale larger than human-to-human economic activity. The video says developers may need to provide clear APIs, authentication, spending limits, and explicit permissions.
• Trust and accountability infrastructure: Looking further ahead, Kelly imagines specialized AIs operating throughout everyday life, requiring ways to verify identities, establish trust, prevent fraud, and determine responsibility when systems fail.
• Uncertain progress: AI could continue improving for another 10–15 years without settling debates about AGI, employment, or whether the investment was worthwhile. The video cautions that past accurate predictions do not guarantee future accuracy.

Sponsor Demonstration: Supabase

The narrator demonstrates a simple feedback form backed by Supabase:

1. Create a feedback table in a PostgreSQL database with a title, comment, and timestamp.
2. Connect a form directly to Supabase so its JavaScript client inserts submissions as database rows.
3. Use row-level security to allow submissions without letting visitors read other users’ feedback.
4. Add a database webhook that triggers an Edge Function to send a notification email when a new row is added.

Main Speakers and Sources

• Main speaker: The video’s narrator, who evaluates Kelly’s predictions and provides commentary.
• Kevin Kelly: The futurist and author whose earlier predictions and newer essays are discussed and quoted.
• Other cited source: Warren Buffett, referenced through a quotation about forecasts revealing more about forecasters than the future.