Brendan Babb’s Post

“Let’s work this out step-by-step.” Have you wondered why so many people are adding this part when interacting with a LLM? Since last year, this has become pretty much a popular way of prompting, after showing its clear effects in terms of accuracy and efficiency. But do you know why? This approach isn't just about getting straight answers from the machine. Instead, it encourages LLMs to walk us through their thought process (here’s the popular “step-by-step”), before landing on a solution. The difference? Instead of an AI spitting out answers with no context, we get to see the how and why behind its conclusions. Here are a couple of examples: —> You are planning a road trip from City A to City B, which is 300 miles apart. If you drive at an average speed of 60 miles per hour, stopping for a total of 1 hour for breaks, calculate the total travel time. Please provide a step-by-step explanation of your process. —> You're an environmental scientist analyzing the impact of reducing plastic use in a small town. The town consumes 1 million plastic bottles a year, which they plan to reduce by 20% by promoting reusable bottles. Walk me through the environmental benefits of this reduction. Especially when dealing with more complex tasks, the Chain-of-Thought prompting technique is a great alley to dig deeper into the process! #GenerativeAI #BasicsOfPrompting #ChainOfThought

Computer vision-based sorting is essential in countless industries, for example, food processing, manufacturing, and waste/recycling industries. The ability to accurately train a machine learning model to properly identify all the unique objects as well as all different orientations that the objects may be in requires vast amounts of annotated data. If you were to capture the amount of necessary real-world data required to train your machine learning model, it would be extremely time consuming and cost prohibitive to get the domain randomization needed. Additionally, you would be required to manually annotate all the real-world data.   CEBS offers Synthetic Data as the solution for this problem. We are able to generate the datasets that you need to train your computer vision-based sorting machine learning models. Synthetic data comes with programmatic annotations as well as domain randomization to include object sizing, positioning, rotation, camera positioning and rotation, lighting variations, wear and tear of objects and surrounding components, and much more.   Reach out to us for your solution to data scarcity problems in training your computer vision machine learning models. Check out our recycling sorting synthetic data example that we demoed at CVPR 2024.   #syntheticdata #syntheticdatageneration #AI #ML #artificialintelligence #machinelearning #computervision https://lnkd.in/gdN4XP4j

🔥 A few months ago, our COO and Co-Founder, Drake H., sat down with a panel of industry leaders at #ISRI2024 to discuss the impact of AI in the recycled materials industry. Missed the presentation IRL? We've got you covered. ✔️ Check out the fascinating conversation between Drake, Sid G., Josh Singer, and Dennis Xu, and let us know - how do you see AI shaping the industry in the future? 🤔 #AI #ArtificialIntelligence #recycling #scrapyard #software

The deeper I go into AI the more excited I become about the potential. In this video we see how Google DeepMind #alphafold is used to help find enzymes which can break down plastics for recycling. #artificialintelligence #proteinfolding #circulareconomy #reducewaste #innovation #recycling

#recycling #plasticsorting #meyersorter What can Meyer do for the resource recycling industry? Meyer has the industry's most advanced sorting technology, enabling sorting from whole bottles to bottle flakes, pellets, etc. Equipped with AI deep learning innovation, the efficiency and sorting speed are outstanding. It is mighty in identifying hard-to-classify objects, leading to significantly higher purity results. Choose Meyer to realize sorting whatever you want! #recycling #plasticsorting #meyersorter

That's a lot of plastic garbage that will take a long time to break down. This is just one food order for my lunch at work. Is this a consumer problem, policy problem, or technology issue? AI won't solve this just like AI won't solve the real problems we're experiencing in healthcare. We have all the necessary technology for my food to arrive in proper containers with minimal container usage. It could be a consumer issue. If I just stopped ordering food in then perhaps the problem would be solved. Though history would tell us that's not the case. That leaves a policy issue.

Opt for environmentally friendly packaging materials like recycled paper to reduce plastic waste and meet sustainability goals. #Sustainability #EcoFriendly #RecycledPackaging #ReducePlasticWaste #EnvironmentallyFriendly Learn more about this topic here: https://lnkd.in/efepXp3u At Grow Scale Win™, we're all about using AI to enhance productivity and efficiency. That's why we've built the AI Growth System™, a truly holistic approach to growing your business. You get a personalized growth strategy and a full suite of AI powered tools to generate leads, nurture prospects, and close sales. Learn more at www.growscalewin.com. DM us with any questions.

Caught in the act? Then, $150 fine! This #AI program automatically sends fines to motorists who throw rubbish on the UK roads. Will you implement it in your city? Credit: LitterCam and Waste-Ed #ai #machinelearning #innovation #tech #sustainability #sustech #wastemanagement #council #governmentinitiatives #litter #rubbish

We've just launched the full version of Improvability AI - which is our new brand evolution from OpenESG. Think AI for Business - with more accurate data results than Perplexity and ChatGPT - initially trained on sustainability, ESG and Supply Chain models. Users can build their AI knowledge library and then generate reports, compliance docs, sales and marketing material and training support - watch the video below to see how we compare with Perplexity, ChatGPT. Your library also doesn't disappear after 30 days or when you log out. You can upload large files, including CSV, PPT, PDF etc. The library is there with you as you build it out and really start to utilize AI in your workflows and our specialist AI teams can help companies make the transition to using AI in business with confidence, developing specific solutions for your needs. It's been 12 months of great work across the product team led by Daniel Ellison to get here driven by our visionary CEO Tee Ganbold 🌏 bootstrapping the development in conjunction with the awesome teams at Phaze Ventures and PhazeRo - and we're very excited by the future that 2024 holds - which really feels like the year that AI grows up into enterprise deployments. #enterprise #ai #improvability Ruma Bose Tee Ganbold 🌏 David Aikman Daniel Ellison Valerie Keller Gerard Grech Brent Hoberman David Rowan Azeem Azhar Henrik Torstensson Jay McCarthy Malcolm Nutt Masoud Al - Rawahi Mohammed Al-Wahaibi Abdullah Al-Shaksy Ted Merz, CFA Lance Gould Alessa Berg https://lnkd.in/dhbWciTZ

How can AI complement traditional sorting systems? 💡 TOMRA’s traditional sorting systems, such as the AUTOSORT™, are excellent at differentiating materials. But with materials that are visually very similar, things get a little tricky. That's where AI comes in, with the power of Deep Learning. GAINnext™, the Deep Learning-based add-on to our AUTOSORT™, is the next step in TOMRA's continued investment in AI sorting. Here's what it can do: 🔵 Large scale separation of food grade from non-food grade plastics for PET, PP and HDPE. 🔵 Achieve purity levels over 95%, the most accurate solution available on the market today. 🔵 Drive material circularity with bottle-to-bottle recycling. 🔵 Provide an ecosystem of sorting applications such as deinking and PET cleaner, in addition to the already established sorting of wood chips and PE silicon cartridges. There is more to AI with GAINnext™ and our research and development won’t stop here!