Text Compression as a Test for Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

LLaMA: Open and Efficient Foundation Language Models (2023, background): “Because Turing (2009) proposed to measure machine intelligence by using language through the “imitation game”, language modeling has been proposed as a benchmark to measure progress toward artificial intelligence (Mahoney, 1999).”
Statistical Language Models Based on Neural Networks (2012, background): “Mahoney together with M. Hutter also organize a competition with the aim to reach the best possible compression results on a given data set (mostly containing wikipedia text), known as a Hutter prize competition.”
Recurrent neural network based language model (2010, background): “Sequential data prediction is considered by many as a key problem in machine learning and artificial intelligence (see for example [1]).”
Triple-to-Text: Converting RDF Triples into High-Quality Natural Languages via Optimizing an Inverse KL Divergence (2019, background): “[39] Given a language sample X , its quality is scored by P(X ).”
Measuring universal intelligence: Towards an anytime intelligence test (2010, methodology): “[41], whose Large Text Compression Benchmark eventually converged with Jim Bowery’s CPrize into the Hutter’s Prize9).”
A Minimum Relative Entropy Principle for Learning and Acting (2008, background): “In particular, it has been even proposed that compression ratio is an objective quantitative measure of intelligence (Mahoney, 1999).”