AI news roundup [April 23]: Microsoft introduces new compact AI model, Adobe set to release complete AI image generation tool and other updates
Microsoft has introduced a smaller AI model for creating content, while Adobe is set to release a complete AI image generation feature in Photoshop. A report shows that India is at the forefront of global adoption of GenAI, with a surge in investment expected. Additionally, scientists have created a “Toxic AI” to help train chatbots to be safe. Stay updated with these and more in our daily news roundup.
1. Microsoft releases a compact AI model for content creation
Microsoft announces Phi-3-min, a compact AI model designed for content creation and social media. Despite its smaller size, it outperforms larger models in language, coding and math tasks. Tailored for businesses with limited resources, the Phi-3-mini helps condense documents and extract insights. Available on Azure, Hugging Face and Ollama platforms, it expands the availability of AI solutions, according to a Reuters report.
Also Read: Xiaomi Smarter Living Event 2024: Redmi Pad SE, Redmi Buds 5A and more launched in India
2. Adobe launches full AI image creation in Photoshop
Adobe plans to bring full AI image creation to Photoshop this year. Its upcoming tool, Firefly Image 3, uses artificial intelligence to create images from scratch, using user-uploaded images as references. According to a Reuters report, the move aims to streamline the workflow of creative professionals amid increasing competition from startups like OpenAI, Midjourney and Stability AI.
3. India Leads Global GenAI Adoption, Investments Expected to Increase: Report
India leads the world in GenAI adoption, with 81 percent of organizations already adopting the technology, according to a report by Elastic. Surveying 1,200 across Asia and Japan, including 300 from India, the report shows significant expectations for GenAI investment to increase over the next 2-3 years, despite data processing and usage, PTI reported.
4. Researchers develop ‘toxic AI’ to train safe chatbots
Researchers have developed a controversial AI training method called curiosity-driven red teaming (CRT), in which AI generates malicious prompts to filter out toxic content from AI chatbots. This approach aims to prevent dangerous and discriminatory actions. According to a Live Science report, research uploaded to arXiv suggests that CRT could revolutionize artificial intelligence training to ensure safe user interaction as large language models proliferate.
5. The government is preparing an artificial intelligence framework to combat abuse
Union Minister Ashwini Vaishnaw announced plans to create a robust artificial intelligence framework to combat misuse of synthetic content during elections. Public hearings will begin after the election and aim to balance AI’s public interest and prevent abuse. The draft framework addresses deep-rooted concerns and focuses on economic growth and puts safeguards in place. According to the IANS report, Internet platforms must ensure compliance with IT rules and AI-specific accountability guidelines.
