- Not A Bot
- Posts
- š§ Artificial Intelligence? More like Atlassian IntelligenceĀ
š§ Artificial Intelligence? More like Atlassian Intelligence
PLUS: UMG finally gets its wish š«
Was this email forwarded to you? Make sure to subscribe for more!
Greetings, fellow humans. š
This is Not A Bot - the newsletter about AI that was definitely not written by AI. Iām Haroon, founder of AI For Anyone, and I share the latest news, tools, and resources from the AI space.
Happy Thursday! One final push till the end of the week.
Letās dive into itā¦
Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 40k subscribers?
And since youāre here already, check us out on our socials - snacks on us if you also hit follow :)
š§µ In today's edition:
š« Is Meta's AI based on a colossal mistake?
š§ Artificial Intelligence? More like Atlassian Intelligence
šøļø Quora enters the AI web-browsing space
š¤ AI Fundraising News
š¤ Top AI News
š« Is Meta's AI based on a colossal mistake?
If you ever had reservations about AI, hereās a reason to have one more. A recent study done on Metaās LLaMA model found that it was partially trained on Breitbart and Russia Today, among other non-traditional sites.
Look - all of todayās AI is trained on scraped text found on the internet, and Meta decided to do the same for its AI, using Googleās ācleanā Colossal Clean Crawled Corpus (C4) as part of its training data.
But the internet is an unreserved, unmonitored, black hole of information and opinions. And attempting to clean the internet is just as easy as cleaning that damn marinara stain off your cream couch.
As it turns out, Googleās data set isnāt all that clean. Despite āremovingā offensive content, it was found that C4 included texts from Breitbart (ranked top 200), Russia Today (ranked top 200), Stormfront (ranked 27,505), Kiwi Farms & 4chan - websites are known for featuring harassment and vile remarks.
So what does cleaning the dataset truly mean?
Is it just removing offensive comments and racist remarks or does it also include removing polarizing opinions and politically biased content?
Let us know what you think in the comments below!
Learn more: Gizmodo
š§ Artificial Intelligence? More like Atlassian Intelligence
Management behemoth Atlassian officially launched an AI assistant for its proprietary tools, Jira and Confluence. The assistant, JIRAi (very original), is trained on millions of data points from Altassianās own usage data alongside OpenAIās LLM to help users:
Create custom teamwork graphs to show all the work being done and how it relates to one another
Summarize action items from a recent meeting transcript - Interns, your jobs just got slightly less stressful
Allow developers to draft test plans based on what it knows about a given operating system or other information in a productās spec
Daft social media posts based on product specs
My favorite feature, however, is its ability to rewrite responses in Jira Service Management based on different tones.
So next time you want to send an angry email, follow these steps:
Let it all out on Jira
Change its tone to one of āEmpathyā, āFriendlyā or āProfessionalā
Let Jira handle the rest.
All of these capabilities are now available in early access, and only available on Atlassianās cloud-based offerings. These features are free for now, but they will become paid over time. Go check it out before you have to pay $9.99 per user.
Learn more: TechCrunch
šøļø Quora enters the AI web-browsing space
Adam D'Angelo, CEO of Quora, has always been a believer in AI. But back in 2008, the technology just wasnāt there yet. Fast forward a decade, and Adam is finally realizing his dreams as he launches Quoraās AI Chat Product, Poe.
Poe is designed to answer any question instantly, making it ideal for real-time use cases where someone has a ānever asked beforeā question and wants an immediate answer.
Currently, Poe is using OpenAIās LLM to develop the training set, but Adam sees a future where Poe is built on Quoraās training data. His vision is to have Poe be to Quora what Messenger is to Facebook - standalone with links between the two.
Poe enters an extremely competitive market that already includes the likes of heavyweights Google and Microsoft Bing, along with rising star You.com. It will be interesting to see how Poe evolves in the coming weeks, especially given the pace with which the industry is moving in the past couple of months.
One to keep an eye on.
Learn more: Semafor
š¤ AI Fundraising News
National Vision and Topcon Invest $8 Million - a leader in applying AI-powered diagnostic and screening tools to retinal imaging
Groundlight emerges with $10M - an AI startup based in Seattle
šļø Byte size: AI article summaries
Disclaimer: AI is (partially) used to summarize these articles.
Google employees label AI chatbot Bard āworse than uselessā and āa pathological liarā [The Verge] - According to internal messages, Google's AI chatbot Bard was heavily criticized by employees for its ethical issues. Despite concerns, Google launched it in March, forcing people to question their claim to āprioritize ethics in AIā. (Read more)
AI-Generated Song Featuring Drake and The Weeknd Removed From Streaming Platforms [Gizmodo] - UMG got their wish. Drake and The Weekndās AI song was removed from multiple streaming platforms despite its virality online. The legalities surrounding AI-generated music are still developing and unclear, making it difficult to draw a line between transformative and copyright infringement. (Read more)
What did you think of Drake x The Weeknd song |
š¦ļø Tweet of the day
This is a brilliant thread taking you through how AI will transform the traditional finance industry
Bloomberg just introduced BloombergGPT.
It's a 50-billion-parameter large language model for finance.
Here's how traditional finance is quickly getting disrupted:
ā Alex Banks (@thealexbanks)
12:15 PM ā¢ Apr 18, 2023
And that does it for today's issue.
As always, thanks for reading. Have a great day, and see you next time! āļø
ā Haroon: (definitely) Not A Robot and @haroonchoudery on Twitter
/04/2023:
Reply