This episode of Hashtag Trending is brought to you by CDW Canada Tech Talks. If you’re passionate about technology and innovation, this is the podcast for you.Join host KJ Burke, as he and industry experts dive into the latest trends, insights, and strategies shaping the tech landscape in Canada. Visit cdw.ca/tech talks to tune in today. There’s a link in the show notes.
HTML May Be A Solution To AI Hallucinations, Privacy-First Mobile Carrier Launches for High-Risk Users Perplexity’s AI Shopping Assistant Plans To Revolutionize E-Commerce
Welcome to Hashtag Trending. I’m your host, Jim Love. Let’s get into it.
California-based conversational search engine Perplexity is transforming online shopping with its new AI-powered assistant. The platform now allows users to research and purchase products seamlessly, competing with major players like Google, Amazon, and Walmart.
Perplexity doesn’t have its own AI. Some have called it a “wrapper” around current AI. As such, it has to deliver real value, especially as rivals such as OpenAI enter the AI search space. This new shopping feature may be the ticket.
Perplexity Pro users in the U.S. can make purchases directly through the platform with one-click checkout, eliminating the need for third-party merchant visits. Shipping is free for all orders made with this feature.
A visual search tool lets users find products by simply taking a photo, making it easier to locate items without needing detailed descriptions.
And finally, Perplexity provides tailored, ad-free product recommendations based on user queries, using integrations with Shopify for up-to-date product information.
Competing in a Crowded Market
CEO Aravind Srinivas emphasized that Perplexity aims to provide a clean, user-focused shopping experience without ads or spam, setting it apart from competitors like Amazon’s Rufus assistant and Walmart’s generative AI tools. He added, “We’re on it” to make shopping more magical by focusing on simplicity and user intent.
This shopping feature builds on Perplexity’s expanding toolkit, which includes live finance analysis tools, real-time election information hubs, and advanced reasoning modes. These innovations have earned praise from industry leaders like NVIDIA CEO Jensen Huang, who uses Perplexity “almost every day.”
Expansion and Challenges
Perplexity is reportedly raising funds at a $9 billion valuation to fuel its global expansion and navigate legal challenges. Recent plagiarism lawsuits from major publishers like News Corp have increased financial pressures, though the company has also struck revenue-sharing deals with outlets like TIME and Der Spiegel. Srinivas has called for collaboration between media and tech to create innovative tools that benefit all parties.
By integrating AI seamlessly into e-commerce and user experiences, Perplexity may challenge industry giants and lead the evolution of online shopping.
Researchers Propose HTML-Based Solution to LLM Hallucinations
A team of Chinese researchers has introduced *HtmlRAG*, a novel method to enhance large language models (LLMs) by incorporating structured HTML data into Retrieval-Augmented Generation (RAG) systems. The approach could mitigate hallucinations—a common problem in AI-driven tools like ChatGPT and Perplexity—by preserving the semantic and structural richness of web content.
Traditional RAG systems often strip HTML data down to plain text, losing key elements such as headings and tables. This compromises the contextual depth of retrieved information. *HtmlRAG* retains HTML’s structure using a two-step pruning algorithm. The first step cleans unnecessary HTML elements, reducing content length by 94%. The second uses a block-tree-based pruning method, combining embedding and generative techniques to retain only relevant content, ensuring compact yet rich inputs for LLMs.
While effective, the HTML-based approach introduces complexities, including handling additional tags, JavaScript, and CSS. Some experts, like Cristina Belderrain, advocate for Markdown as a simpler alternative. However, proponents of *HtmlRAG* argue that HTML’s widespread use and ability to capture detailed structure make it more versatile for web-based knowledge retrieval.
This method builds on advancements like Google’s Vertex AI and Microsoft’s Azure AI, which also integrate HTML processing. Unlike these systems, *HtmlRAG* directly feeds HTML structure into the pipeline, preserving critical relationships and enabling more precise and efficient knowledge integration.
Biden Secures $6.6 Billion TSMC Deal to Lock in CHIPS Act Legacy
President Joe Biden’s administration has finalized a $6.6 billion award to Taiwan Semiconductor Manufacturing Company (TSMC) to build chip factories in Phoenix, Arizona. The deal comes just weeks after Donald Trump’s election victory, signaling an effort to cement the CHIPS and Science Act’s impact before a potential policy shift under the new administration.
The funding supports TSMC’s $65 billion investment to establish three state-of-the-art semiconductor fabrication plants in Arizona. The facilities will produce advanced chips for major U.S. technology firms like Apple, NVIDIA, and Qualcomm, creating thousands of jobs and bolstering the U.S. semiconductor supply chain. The first plant is expected to become operational within months.
“For the first time in decades, an American manufacturing plant will be producing the leading-edge chips used in our most advanced technologies,” Biden said. The CHIPS Act, signed into law in 2022, allocated $52.7 billion for semiconductor research, manufacturing, and workforce development to reduce reliance on foreign suppliers.
Trump, however, has criticized the act, calling it “a bad deal” and accusing Taiwan of dominating the U.S. chip industry. Despite his opposition, TSMC’s investment underscores bipartisan recognition of the strategic importance of domestic chip production to national security and economic stability.
Privacy-First Mobile Carrier Launches for High-Risk Users
A new privacy-focused telecom service, Cape, is launching today, offering public figures, executives, journalists, and activists a secure way to use mobile networks with minimal data collection. The service aims to address rising concerns over government surveillance and cyberattacks, particularly from state-sponsored actors.
Cape’s service, built on the U.S. Cellular network, uses proprietary software to limit user data collection. Unlike traditional carriers, Cape stores only essential subscriber information for 60 days and enables users to rotate identifiers like device and advertiser IDs on-demand. This makes it harder for apps and brokers to track and monetize user data.
Cape’s device, a standard Android phone with modified data settings, is optional. Customers can use their own phones and port their numbers, with updates handled via eSIM technology. CEO John Doyle, a former Palantir executive and U.S. Army special forces sergeant, explained that Cape tested its services with national security professionals, government officials, and privacy advocates to refine its offering.
While the service provides robust privacy protections, it comes at a high cost, with some plans reaching $1,000 monthly. Cape is working to make its service more affordable for groups like journalists and domestic violence survivors and plans to expand to the general public next year with competitive pricing.
And that’s our show for today.
Thanks to our sponsor, CDW and KJ Burke’s CDW Canada Tech Talks. Check it out if you get the chance. You can find it like us on Spotify, Apple or wherever you get your podcasts.
Reach me at editorial@technewsday.ca
I’m your host Jim Love, have a Fabulous Friday
