🙋🏽♀️ Notion has questions (and answers)
Plus Microsoft announces its own AI chip
Login
Productivity app Notion is rolling out an AI-powered Q&A feature, allowing users to ask questions about content in their workspace.
Users are presented with a text box where they post their questions, and the tool compiles a response based on the data within the workspace. The Q&A function also works across shared workspaces.
Notion Q&A feature is currently in limited availability (you have to join a waitlist), but it will eventually be offered as part of its AI service, which costs $8 a month for individuals, or $10 a month per user for teams.
**Why it matters: **AI chatbots are good for general purpose use, but companies are starting to see value in letting users mine their own data; indeed, **Microsoft **and **Google **have already hopped into this space with Copilot and NotebookLM respectively. It will be interesting to see how Notion’s Q&A tool fares against these 300 lb gorillas.
Microsoft is getting into the AI chip game.
The tech giant has announced Maia 100, a chip specifically designed to run cloud AI workloads. Microsoft is testing Maia 100 with **OpenAI’s **GPT-3.5 Turbo model, as well as its proprietary Bing AI, and GitHub Copilot.
Maia 100 makes use of a 5nm manufacturing process from chip heavyweight TSMC, and promises to be competitive with Nvidia’s H100 GPUs.
Maia 100 will be made available via Azure, though Microsoft is yet to announce a launch date.
**Why it matters: **OpenAI previously relied upon expensive H100s from Nvidia for training its GPT models, but one would expect it will rely on Microsoft from here on out - especially as it’s leaning on the company for even more money. But with Nvidia’s next-gen H200 just around the corner, it remains to be seen how Microsoft’s Maia 100 will hold up.
**YouTube **has announced new rules that will require creators to disclose whether their content is AI-generated, or incorporates AI elements in any way.
The platform, which is owned by Google, says the new terms will roll out in the coming months, and notes creators who do not abide could find themselves demonetized or even have their accounts suspended.
Additionally, YouTube will soon let viewers flag videos for potential removal if they believe AI has been used to imitate someone.
Why it matters: The power of AI to create art is well-known, but with the rise of deepfakes, the potential for harm is also coming into sharp relief. However, YouTube is a massive platform with an estimated 500 hours of content uploaded every minute, which will likely make it significantly more challenging to police AI videos compared to, say, copyrighted music.
Health-tech startup Forward has raised $100 million for its AI-centric walk-in health pods.
Dubbed CarePod, the eight feet by eight feet kiosks are designed to be placed in malls, where patients can enter for an AI-powered diagnostics procedure such as a blood test or body scan.
It’s of note that, in the US, only licensed healthcare professionals can make a health diagnosis or write a prescription. Still, Forward founder Adrian Aoun suggests the CarePod could one day take over such responsibilities: “Slowly but surely we're just migrating every single thing from doctor and nurse to hardware and software,” notes Aoun. “In fact, we don't even believe a doctor's office should exist.”
Forward charges users $99 a month to use CarePods, which are currently spread across a handful of states and Washington DC.
**Why It matters: **There are a lot of potential benefits when it comes to AI and the healthcare space, but getting it right will be vital. After all, the Theranos scandal serves as a looming reminder of what can happen when we put too much faith in health-tech.
