Why generative AI represents a new paradigm in computing

Anyone familiar with a British comedy show called Little Britain might remember the catchphrase ‘computer says no’. The sketch epitomised humanity’s early relationship with technology. Computers tended to be a limitation (or source of errors). Jobsworth operators that couldn’t see past those limitations would cause no end of frustration for their fellow humans.

Generative artificial intelligence introduces a new paradigm. Computers are no longer limited to binary outcomes or a drop-down list of options. They can now answer questions and communicate in very human-like ways. In fact, they can be a good deal more articulate and knowledgeable than most humans.

OpenAI’s GPT-4 outperforms the average human[1] at a range of tasks. It passes SATs (standardised high school level tests) with flying colours – in the 93rd percentile for reading and writing, and 89th percentile for maths. From there, it goes on to do very well in its chosen vocation – passing the bar exam in the 90% percentile and the medical knowledge self-assessment program in the 75% percentile.

Think about what that means: GPT-4 is already better than 90% of prospective lawyers at answering questions about the law. A ‘general purpose’ computer beats 90% of highly intelligent, well-educated human beings at their specialist subject. Given that the technology is in its relative infancy, it’s fair to assume that it won’t be long before it outperforms even the smartest human.

Outperforming humans

AI outperforming humans isn’t a novelty. Deep Blue first beat world chess champion Gary Kasparov more than 25 years ago[2]. And it’s been more than seven years since AlphaGo beat world Go champion Lee Sedol[3]. But these victories for computer-kind were fought in a confined problem space, with a relatively trivial impact on society. Who cared if computers were better than us at chess, or at Go? We still had the edge in the things that make us human.

Generative AI is important (and scary in equal measures) because it encroaches on attributes that we long considered our exclusive forte – mastery of language, creative writing, knowledge-based work, writing software code… even art and music. That means its potential applications and societal implications (for better or worse) are staggering.

[1] https://openai.com/research/gpt-4

[2] https://en.wikipedia.org/wiki/Deep_Blue_versus_Garry_Kasparov

[3] en.wikipedia.org/wiki/AlphaGo_versus_Lee_Sedol

This article- entitled “Why generative AI represents a new paradigm in computing” is the first in a seven-part series. Others articles within the series include:

Why generative AI represents a new paradigm in computing

The risks of implementing generative AI

Mitigating the risks of generative AI

The economic upside of generative AI

The downside of generative AI

Does AI pose an existential threat to humanity? 

How will AI change the future?