Weekly Tech Recap: Google AI search tells users to eat rocks, OpenAI vs Scarlett Johansson and more – Crypto News

Top tech news of the week:

1) Scarlett Johansson accuses OpenAI of imitating her voice: 

Earlier this week, Avengers actor Scarlett Johansson alleged that OpenAI copied her voice for ChatGPT‘s ‘Sky’ system. Reacting to the allegations, the Sam Altman-led company has taken down the ChatGPT voice in question while also stating that Sky system’s voice is not an imitation of Johansson but belongs to a different professional actress.

Johansson stated that OpenAI CEO Sam Altman had first approached her in September 2023 for being the voice of ChatGPT. However, the actor stated that despite her rejecting the offer, Altman went on to pursue a voice for ‘ChatGPT 4.0 system’ that sounded just like her. 

“When I heard the released demo, I was shocked, angered and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference.” Johannsson stated in her statement. 

2) Google AI search tells users to eat rocks: 

Months after Gemini‘s text-to-image feature caused a major controversy, another Google AI feature has landed the company in hot water. Google Search’s AI overviews feature, which made its way into the stable version of Google Search earlier this month, was seen making a number of questionable suggestions to users.

In response to one query, AI Search, citing UC Berkeley scientists, said it was recommended to eat “at least one small rock a day” because they contain minerals and vitamins that are important for digestive health.

In response to a query about ‘cheese not sticking to pizza’, the AI overviews feature suggested that users add ‘non-toxic glue’ to their pizza to give the sauce more ‘tackiness’. What’s even more shocking than this answer is the fact that the feature relied on an 11-year-old Reddit post as its source.

In a statement given to 9to5Google, the company said that a ‘vast majority of AI Overviews provide high quality information’ and it has ‘seen examples that were doctored or that we couldn’t reproduce’. 

“We’re taking swift action where appropriate under our content policies, and using these examples to develop broader improvements to our systems, some of which have already started to roll out.” Google stated in its statement. 

3) Truecaller’s AI-powered feature can replicate your voice: 

Truecaller launched a new AI-powered ‘personal voice’ feature that will allow users to create a digital version of their own voice. This voice can be used to greet users and identify the purpose of their call. The new feature, part of Truecaller’s AI assistant, is powered by Microsoft Azure AI speech technology.

Truecaller’s AI assistant has been part of the application for some time. It can handle tasks such as answering and screening calls, taking messages, replying on behalf of the user and recording the conversation. However, with the new feature, Truecaller will allow users to have their calls answered in their ‘real’ voice, rather than in a synthetic-sounding voice by a digital assistant.

Truecaller says the new feature will create a more engaging experience for the caller when interacting with the virtual assistant. To use the feature, users must record their voice for a few seconds while the AI assistant is set up.

4) Microsoft launches AI PC’s, team Copilot at Build 2024 event: 

Microsoft has taken the competition to Apple with the introduction of new Copilot AI PCs, powered by the new ARM-based Snapdragon X Elite chipsets to deliver 58% more performance and 20% longer battery life than the MacBook Air with the M3 chipset.

The Satya Nadella-led company also unveiled a new version of Copilot, called Team Copilot, designed to improve group collaboration. Team Copilot will act as a meeting assistant, able to take notes, summarise key points and track deadlines.

5) WhatsApp could bring a new AI-backed feature to create profile pictures: 

A report from WABetaInfo earlier this week revealed that WhatsApp could soon introduce an AI-powered feature to help users create their profile pictures. According to a shared screenshot, users will be able to enter text prompts into an AI image generator, which will then create a personalised profile picture based on the details given by the user.

The report doesn’t highlight which AI model Meta is using to develop this feature. There’s also no confirmed timeline for when the new feature might make its way into the stable version of the WhatsApp application.

Milestone Alert!
Livemint tops charts as the fastest growing news website in the world 🌏 Click here to know more.

You are on Mint! India’s #1 news destination (Source: Press Gazette). To learn more about our business coverage and market insights Click Here!

Published: 25 May 2024, 04:55 PM IST