- AI Now
- Posts
- Is AI Coming For Our Jobs?
Is AI Coming For Our Jobs?
Plus: AI powered email inbox, idea-to-design, a big announcement & more!
Happy Monday and welcome back to AINow. Before we get to the news we have an exciting announcement to kick off this wonderful week.We are thrilled to announce that we are taking AI education to the next level! We know that there are countless AI tools on the market, each claiming to be the best and each costing seemingly $20 a month. This can be overwhelming and can also add up quickly. But don't worry, we've got you covered!
Starting soon, we will be giving our subscribers access to the latest and greatest AI tools, all for free! That's right, you heard it correctly, no more shelling out money every month to try out new AI tools. We want to make sure you get the most out of these tools and that's why we'll be providing you with insights and tutorials on how to use each tool to its full potential.
Here's how it works: At the start of each week, you will vote on the tool you want to try out and learn how to use. Based on the votes, we will secure free access for our subscribers. Then, throughout the rest of the week, we will be providing in-depth tutorials and tips on how to best use the tool so you can decide if it is a tool you want to add to your toolbox.
Stay tuned for more information on this in the coming issues this week. We can't wait to help you get started on your AI journey and find the tools that will take your projects and daily tasks to the next level. Now, lets get into today's issue.
📍 Our top picks
The Canadian federal government will invest $39.3 million in over a dozen artificial intelligence projects across various sectors such as manufacturing, retail, and aeronautics. The funding will be invested through Scale AI, a Montreal-based technology supercluster funded by the federal and Quebec governments.
Private investors are also participating in the latest financing round, bringing the total investment to $117 million across 15 different projects.
Artificial intelligence (AI) and automation technology are changing the job market but not necessarily eliminating jobs. The perception that robots are taking jobs is fueled by videos of "fully automated" restaurants, but these are often misleading as they still require human workers in the back.
AI customer service bots still require human backups for complex situations and many services outsourced human work to cheaper labor markets. The fear of robots taking jobs has been around for less than a decade and recent studies suggest that while 85 million jobs may be replaced by machines by 2025, 97 million new jobs will be created to support the new economy.
🔧 AI Toolbox
Idea to design in an instant. Create delightful, editable UI designs from a simple text description.
AI powered email inbox that automatically sorts important emails based on behavioral patterns.
Allows you to set reminders and actually holds you to them. Equipped with a open rate tracker and undo send feature
Create a heartfelt Valentine's Day card for your loved one in just a few seconds
📊 News
🔑 Use Case
🧠 Learn
What is RT-1?
A massive neural network called RT-1 that Google created can be used to command actual robots in the real world. RT-1 is essentially an effort to build a sizable pre-trained model that incorporates the experiences of various robots performing various tasks into a single model and uses this model to operate actual robots in the real world. Preliminarily, the strategy appears to be effective, but don't get too happy because there is always a huge gap between "sort of works" and "put this in a product and sell it to a grandmother."
What is RT-1? Over the course of 17 months, a fleet of 13 robots used at Google collected 130k episodes of robot behavior covering 700+ tasks, which were used to train RT-1. In comparison to earlier methodologies, "we demonstrate that RT-1 can display much greater zero-shot generalization to novel tasks, settings, and objects," says Google.
Compounding Returns: RT-1 can be combined with other methods to improve robot performance in practical settings. For instance, Google utilized RT-1 to program robots connected to SayCan to behave in certain ways (a system that uses a large language model for helping the robot to plan actions - see Import AI 291). They write: "SayCan with RT-1 achieves a 67% execution success rate in Kitchen1, outperforming other baselines (up from 47% for just vanilla SayCan) while SayCan with Gato and SayCan with BCZ shapely fall in performance due to the generalization difficulty presented by the new unseen kitchen, while RT-1 does not show a visible drop."
Reply