‘…it’s likely to be a long while before machines can really converse with us. Language is deceptively difficult for machines to make sense of, in part because of its compositional complexity: words can be rearranged to unlock infinite meaning.’
US chip-maker Nvidia has developed software that dramatically speeds-up natural language processing (NLP). MIT TECHNOLOGY REVIEW reports on the double-edge sword effect: uses-for-good include voice assistants that do more and sound more natural. At the same time, bad actors gain more sophisticated ways to deceive users with fake media.
Nvidia optimized speed and data to reach its new NLP plateau:
- Training is quicker – from weeks to days
- Processing is faster – from 40 milliseconds to under 2 milliseconds
- Modelling is with more data– their new model has 8.6 million parameters