OpenAI has responded to its rivals’ attempts to surpass ChatGPT as the king of the large language models, rolling out a small but potentially useful update called Read Aloud.

As the name suggests, Read Aloud allows ChatGPT to read out its responses in the user’s language, with users given a choice of five different voices. It’s designed to make ChatGPT easier to use while on the go, and is available for both Android and iOS, as well as the web-based app.

OpenAI explained that mobile users can activate the new feature by pressing down on ChatGPT’s response to activate a pop-up menu. From there, they’ll be able to choose the narration feature.

ChatGPT can now read responses to you.

On iOS or Android, tap and hold the message and then tap “Read Aloud”. We’ve also started rolling on web – click the “Read Aloud” button below the message. pic.twitter.com/KevIkgAFbG

— OpenAI (@OpenAI) March 4, 2024

OpenAI said it can be accessed on its web app by clicking on a small microphone icon that’s been added to the page, just below the chat window.

The company said in a statement that Read Aloud might “come in handy when using ChatGPT on the go, in moments when you need to be hands-free, or in instances where hearing a message out aloud helps you understand it better.”

Read Aloud adds to the voice chat capabilities that became available in ChatGPT last September, which allows users to ask questions or provide prompts without typing. When that feature launched, ChatGPT would still output its responses only in text.

However, Read Aloud means users can set the chatbot up to respond verbally. By adding the two features together, users can now engage in true conversations with ChatGPT.

OpenAI noted that Read Aloud supports 37 languages at launch, but users won’t have to worry about this because the app auto-detects their language and responds in the same way.

The feature is available with both GPT-4 and GPT-3.5. It comes just hours after one of OpenAI’s biggest rivals, Anthropic PBC, announced what is claimed to be the most powerful LLM ever made.

OpenAI might have an early mover advantage but it’s having to fight tooth and nail to retain its crown as the world’s top generative artificial intelligence provider. Late last year, Google LLC debuted its Gemini family of models, which it claims is more powerful than GPT-4, though it has suffered some negative press on its image generation and come in for some fairly heavy criticism over its safety guardrails.

Earlier today, Anthropic launched its Claude 3 series of generative AI models, and claimed that the most advanced one, known as Claude 3 Opus, outperforms both GPT-4 and Gemini Ultra with its “near-human levels of comprehension and fluency on complex tasks.”

Anthropic offered some benchmarks to back up those claims, and on the GSM8K test that involves solving grade school math problems, Claude 3 Opus answered 95% of questions correctly, compared with 94.4% and 92% for Gemini Ultra and GPT-4, respectively.

Still, OpenAI may soon be able to challenge Claude 3 Opus. In November, the company said it is currently developing a successor to GPT-4 that will bring more advanced capabilities. Google is also working on some “dramatic improvements” for its Gemini series models, in a sign that suggests the race for generative AI dominance has only just begun.

Photo: Levart_Photographer/Unsplash