Google’s AI tool predicts floods in South America and Africa with a four-day early warning period; Breakthrough AI system MinD-vis recreates visual experiences from brain waves; Apple’s iPhone 15 launch and AI developments; Contextual AI teams up with Google Cloud to advance generative AI for enterprises – this and more in our daily roundup. Let us take a look.
1. Google’s AI tool predicts floods
Google’s AI flood prediction system, Flood HubGoogle, has achieved a breakthrough by forecasting floods up to four days in advance, according to a NewsScientist report. This technology, initially successful in data-rich regions like Europe and the US, now extends its reach to data-poor areas, notably South America and Africa. By employing transfer learning, Google’s AI leverages knowledge from well-measured regions to make precise predictions in regions lacking reliable data, addressing a critical challenge in global flood prediction.
2. Breakthrough AI system MinD-Vis recreates visual experiences from brain waves
An AI system called MinD-Vis can interpret brain waves to recreate visual experiences. Singaporean researchers aim to integrate MinD-Vis into VR headsets, enabling mind-controlled interactions in a metaverse. By analysing brain scans of 58 participants exposed to diverse images during MRI scans, MinD-Vis crafts individual AI models, according to a euronews report. This “mind-reading” technology translates brain activity into images, resembling what participants see. Potentially, it holds promise for future applications for assisting people.
3. Apple’s iPhone 15 launch and AI developments: stock uptrend potential
As anticipation builds for the iPhone 15 launch, questions arise about its potential to reignite Apple’s stock rally. According to reports from MarketBeat, Speculation abounds regarding whether AI innovations could hold the key to reversing the stock’s fortunes. Notably, Warren Buffett’s Berkshire Hathaway has a substantial stake in Apple. While Apple is discreet about its generative AI initiatives, reports suggest the development of an AI-driven chatbot, possibly indicating untapped potential for investors, though some may opt for a more cautious approach.
4. Contextual AI teams up with Google Cloud to advance generative AI for enterprises
Contextual AI has formed a strategic partnership with Google Cloud to empower its enterprise-focused generative AI endeavours. Contextual AI, emerging from stealth mode, aims to develop customizable, privacy-conscious AI models for businesses. Google Cloud’s leadership in generative AI and robust AI/ML infrastructure, including GPU VMs and Tensor Processing Units (TPUs), will facilitate the training and scaling of Contextual AI’s large language models, marking a significant collaboration, Cision PR Newswire reported.
5. Do AI models, like GPT, truly understand humour?
At ACL ’23, a “Best Paper Award” winner investigated if AI models like GPT truly comprehend humour. The study, led by Jack Hessel from the Allen Institute for AI (AI2), scientifically examined the ability of large neural networks to grasp jokes. Co-authors, including Ana Marasovic, Jena Hwang, Lillian Lee, Jeff Da, Rowan Zellers, Robert Mankoff, and Yejin Choi, aimed to explore the depth of AI’s ability to understand humor. According to the report, “The researchers discovered that there’s much room for improvement for AI to come close to achieving a human-level understanding of humor.”