Reaction
AGI won't have a single finish-line test, DeepMind SVP believes
Google DeepMind says its SVP sees progress in everyday reliability, not a single AGI benchmark.
TLDR
More dependable models in daily life, rather than one decisive score: that's how Google DeepMind describes its SVP @koraykv's view of progress toward artificial general intelligence (AGI). The company's post highlights his conversation with @fryrsquared, covering products, model architectures, recursive self-improvement, safety and benchmarks.
Combined views
2.5K
2 Sources, first seen ago

Comments