Cambridge researchers test ChatGPT-4 on an ancient Greek maths puzzle, revealing strengths, weaknesses and the importance of guided AI reasoning.
Many turn to ChatGPT to check their symptoms and learn more about health concerns, but for Warren Tierney, an Irish man, the reassurance he received from the AI tool led to a delay in his cancer treatment. Now, he is warning others about the dangers of over-relying on technology, which can sometimes cause more harm than good.
A recent study of Frontiers in Psychology found that ChatGPT-4 outperformed human psychologists in understanding and responding to human emotions.
Some of the biggest names in tech, including Elon Musk, are calling for a pause on training systems more powerful than OpenAI's newly launched model GPT-4.
A research paper pointing out how ChatGPT-4 aced a US bar examination is a sign of things to come. The free ChatGPT3.5 version doesn’t seem so attractive any more, but it is worthwhile appreciating OpenAI’s evolutionary work on large language models that got us here
The term, AI, breeds misunderstanding and helps its creators avoid culpability. Reasonable technologists have tried for years to replace “AI” with “machine learning” but that doesn’t trip off the tongue in quite the same way
Example: ChatGPT can summarise lengthy articles and documents, which may be useful at firms that must assimilate a great deal of information swiftly.
ChatGPT-4’s answers to the how-do-we-fix-Argentina's-inflation question seemed no better than those of the many humans who have given it a shot and failed
It's creepy that a bot would decide to deceive, but perhaps we shouldn't be surprised. ChatGPT-4 have come to understand from texts it has trained on that human beings often use lies to get their way
Don't worry the robots aren't taking over...yet!
OpenAI CEO Sam Altman tweeted 'We love India' as ChatGPT Plus subscriptions became available in the country. The subscription plan will give users early access to premium features.