From writing articles to helping stroke patients: Here are 6 AI updates this month

1. ChatGPT developer closes in on $30bn valuation in talks to raise capital

OpenAI, the artificial intelligence research lab behind chatbot ChatGPT, is in talks to sell existing shares in a tender offer that would value the company at about $29 billion, the Wall Street Journal reported, citing people familiar with the matter.

OpenAI's chatbot is a software application designed to mimic human-like conversation based on user prompts and can respond to a large range of questions while imitating human speaking styles.

The firm expects business to surge as it pitched to investors saying the organization expects $200 million in revenue next year and $1 billion by 2024, Reuters reported in December. Meanwhile, Microsoft is reportedly eyeing a $10 billion investment into OpenAI, Semafor reported this week.

2. AI is improving outcomes for stroke patients

AI has led to a significant improvement in stroke patient recovery, according to the UK's Department of Health and Social Care. The technology is helping reduce the time between diagnosis and treatment, as well as assessing the best options for treatment. It has tripled the percentage of people who recover to be able to perform daily activities from 16% to 48%.

The Brainomix e-Stroke system almost halves the time it takes to diagnose and begin treating a stroke. "Every minute saved during the initial hospital assessment of people with stroke-like symptoms can dramatically improve a patient's chance of leaving hospital in good health," says NHS England Director of Transformation Dr Timothy Ferris.

3. Princeton student builds app to detect ChatGPT-generated text

The release of ChatGPT has sparked major conversations in academia, with many sceptics raising concerns about plagiarism and sourcing. Yet some of that concern may be alleviated after a Princeton student built an app that can detect ChatGPT-generated text.

The app, called GPTZero, aims to help mitigate the issue of "increasing AI plagiarism," the app's creator, Edward Tian, said on Twitter. In a subsequent blog post, Tian said over 10,000 people have tested out an early version of the app.

4. Generative AI platform creates videos from a single-word instruction

There's been a loud buzz around AI that can generate images and written content from simple instructions. ChatGPT, DALL-E and MidJourney were the talk of late 2022. With a new year comes a new platform, and this one can make short videos that could be used on Instagram, Snapchat or TikTok. In fact, QuikVid can generate a video with as little as a single word to guide its creative process, reports TechCrunch.

Tapping in a search for QuikVid now takes you to a waiting list to join the platform. Its creators cite "a huge influx of traffic" as the cause of the hold-ups.

QuickVid’s creator, Daniel Habib, told TechCrunch: “By providing creators with tools to quickly and easily produce quality content, QuickVid helps creators increase their content output, reducing the risk of burnout."

5. Meta's AI CICERO played a world champion at strategy game Diplomacy – this is what happened

Emerging victorious from strategy game Diplomacy requires a combination of "cunning, cleverness, honesty and perfectly timed betrayal" – according to its creator Avalon Hill.

Players attempt to control as much of Europe as possible by making deals and forming alliances with other players.

It's an intellectual challenge for the sharpest human minds, and the game requires strong communication skills. So how would AI get on against a three-times world champion?

To test this, Mark Zuckerberg's Meta developed the CICERO AI tool to play Diplomacy at a high level. In a game that included CICERO and Diplomacy world champion Andrew Goff as competitors, Vice News reports that CICERO learned not to use outright lies as strategy – a tactic the best human players agree with. Click on the video above to hear Goff talk about his experience of playing against a non-human competitor.

6. Fashion designers in Hong Kong showcase collections influenced by new AI assistant

At the Fashion X AI show in Hong Kong, attendees noticed a certain "alien" quality about the new clothes modelled on the event's narrow catwalk – and the designs were, in fact, not entirely human. The show put more than 80 outfits from 14 designers in the spotlight, all of which were created with the help of artificial intelligence software AiDA, short for "AI-based Interactive Design Assistant”. The software was developed by PhD students and academics at the Hong Kong-based AiDLab.

Masked in monochrome blue, wearing outfits that ranged from down jackets to translucent skirts, models strutted past rows of critics and fashion designers.

Attendee Cynthia Tse said it felt like she was witnessing the future of fashion. "I think the face covering is definitely alien-like, and exciting,” said Tse.

According to AiDLab CEO Calvin Wong, the software was created to serve as a “supporting tool” for designers. "AiDA is an assistant for fashion designers just to help them, you know, to work together," Wong said. "Designers and AI can work together to come up with the final collection."


Pemburu Kebenaran