The future of AI is uncertain, with many unanswered questions” : AI, and everything else by Ben Evans: Big Ideas from the report #atomicIdeas

Benedict Evans, a renowned tech industry analyst, explores the trends and potential of Artificial Intelligence (AI) and generative machine learning in his annual presentation at Slush 2023.

He delves into the impact of AI on various industries, the evolution of AI from software to something more, and the uncertainties surrounding Artificial General Intelligence (AGI).

Platform shifts, such as the transition towards generative AI, can significantly alter markets.

Companies are adapting to these shifts by both bundling (integrating generative machine learning into existing products) and unbundling (creating new standalone products).

This process can lead to the creation of new markets native to the new technology.

The concept of Artificial General Intelligence (AGI), a machine’s ability to understand or learn any intellectual task that a human being can, remains an area of uncertainty.

While there’s a split between machine intelligence and general intelligence, it’s acknowledged that true general intelligence is still limited to living beings.

While AI models have shown impressive capabilities in tasks like language processing, they still lack a true understanding of concepts.

The lack of a theoretical model for intelligence and AGI makes it challenging to predict the development of AGI.

Meanwhile, most software companies continue to deploy ideas from the past decade, like SAS, cloud, automation, and collaboration tools.

The broader impact of technology is evident in the widespread adoption of e-commerce and the ubiquity of internet connectivity.

The next phase of technology is not just about what software can do, but how it will integrate into every aspect of people’s lives, similar to how cars became ubiquitous in the second half of the 20th century.

The future of AI is uncertain, with many unanswered questions.

The idea of a universal substrate for building applications, similar to Excel or Notion, is presented as a potential direction for AI development.

However, whether generative AI like chat GPT is a product or just a technology demo remains a question.

The suggestions and ideas generated by AI tools like chat GPT could lead to the creation of new standalone companies.

These suggestions become the basis for unbundling and creating specific vertical applications that go beyond the capabilities of the general-purpose AI tool.

There are different views on how AGI could be achieved – through model size and emergent capability, through producing the right answers without knowing why, or through unknown breakthroughs.

However, the lack of a theoretical model for intelligence and AGI makes it difficult to predict if and when AGI will be developed.

What would you do if you had a million million interns in your office and you could ask them to do stuff? – Benedict Evans

While machines can exhibit superhuman abilities in specific tasks like calculations or memory, true general intelligence is still limited to living beings.

This highlights the difference between having advanced AI capabilities in specific domains versus achieving true artificial general intelligence.

The last 50 years have been about defining what a computer is.

The next 50 years will be about the integration of software into every aspect of life, driven by the widespread use of the internet.

This shift will shape the future of technology, with AI playing a crucial role.