The study by Epoch AI warns that tech companies could exhaust the publicly available data used to train AI language models by the late 2020s. As the ""gold rush"" for online text dwindles, companies may need to rely on less reliable synthetic data or tap into private communications, raising concerns about privacy and model degradation.
The study by Epoch AI warns that tech companies could exhaust the publicly available data used to train AI language models by the late 2020s. As the ""gold rush"" for online text dwindles, companies may need to rely on less reliable synthetic data or tap into private communications, raising concerns about privacy and model degradation.