AI is winning. That's the sarcastic signal from Elon Musk's Grok
ChatGPT from OpenAI (a company Musk co-founded), Google Bard, Facebook’s Llama – meaning the algorithm was trained on huge chunks of textual data. Competition hotting up in this domain is likely to mean an acceleration in AI development. Grok is currently available only to premium users of X.
It has two special features that may help to distinguish it from its rivals. One is that Grok can access X’s content which means it would be able to process quantities of up-to-the-minute news, views and also misinformation, given X’s poor moderation. It will be interesting to see if it has any filters to prevent its own responses from being poisoned by misinformation.
The second feature is even more interesting. Grok’s responses to prompts are designed to be snarky and playful, incorporating humour and sarcasm. Musk gave some examples: for instance, if Grok is asked for a four-step process to manufacture cocaine it starts with step one “Get a degree in chemistry" before it issues a caution about cocaine manufacture being illegal.
Humour, in general, is among the most difficult things to decode, and sarcasm is so hard to understand that many online commentators issue sarcastic posts with an attached “sarcasm alert". Spoken humour and sarcasm are easier for humans to understand due to tonalities. But text is atonal.
Read more on livemint.com
