When it comes to e-commerce giant Amazon, a large number of reviews on the platform might be fake or computer-generated creations, but that’s about to change in a significant way. Spotlighting the capabilities of artificial intelligence, Amazon has announced that they’ll start using generative AI to amplify product reviews. This feature promises to summarize the general sentiment and product capabilities after perusing a substantial array of reviews on a product page. However, it remains to be seen whether this will affect reviewers’ motivation.
The intersection of AI and neuroscience recently got another boost from Berkeley researchers who applied a machine learning model to interpret brain activity while listening to music. Despite the brain being a convoluted network of signals, machine learning has been successful in segmenting identifiable rhythms and melodies.
In the sphere of linguistic models, the Allen Institute for AI has unveiled Dolma, an immense text-based dataset for large language models (LLMs) like OpenAI’s ChatGPT. Dolma is expected to form the foundation for the group’s anticipated open-language model or OLMo.
Turning to the world of entertainment, the recent SIGGRAPH event showcased AI’s proliferating presence as both a tool and a research subject. Despite AI’s reputation for controversy in the film and TV industry, it was utilized innovatively in visual effects (VFX) development, with several great papers on display.
Enhancing the user experience in digital platforms, two apps developed with GPT-4 help visually impaired users. One app can guide the user towards an empty seat in a room, while the other sifts out critical information from medicine bottles. Meanwhile, OpenAI has introduced a moderation technique using GPT-4 to ease the burden on human teams.
Pulling AI into everyday tasks, ETH Zurich researchers developed a method for trained robots to open and walk through doors. Swinging the balance towards user control, a novel tool “Drag Your GAN” allows users to modify only specific parts of an image, potentially revolutionizing the picture editing process.
Amid these advances, there were some slip-ups. Snapchat’s in-app AI chatbot briefly went rogue, causing unease among users, though the company later confirmed it was a glitch. On a positive note, AI startup co-founded by former OpenAI leaders, Anthropic, raised significant funding led by a leading South Korean mobile carrier.
Not to be outdone, a joint endeavor from CMU and Meta created an AI agent that is capable of understanding object interactions and managing tasks, simulating how toddlers learn.
In the end, machine learning continues to infiltrate new frontiers, even solving challenges posed by neuro cells’ unknown role in the brain, piloted by researchers from MIT and Harvard. They propose that certain cells might function similarly to a transformer or attention mechanism.
In conclusion, the state of AI is more exciting than ever, with advancements from Amazon and Google to budding startups and keen researchers continually reshaping the landscape of artificial intelligence with every passing week. Yet, it’s still early in the AI journey, and as with most aspects of this mesmerizing field, only time will tell where we go from here.
