Subterms
<!–> ZDNET’s key takeaways The Nitecore NB Air is is available now for $37. Weighing just 3.14 ounces, it’s the lightest power bank I’ve tested. There’s no USB-C cable or pouch included with the power bank. –> For hikers and runners, every ounce you’re carrying counts. Power banks can be a big challenge because you […] More
Elyse Betters Picaro / ZDNETMost people know that the famous Turing Test, a thought experiment conceived by computer pioneer Alan Turing, is a popular measure of progress in artificial intelligence.Many mistakenly assume, however, that it is proof that machines are actually thinking. The latest research on the Turing Test from scholars at the University of California at San Diego shows that OpenAI’s latest large language model, GPT-4.5, can fool humans into thinking that the AI model is a person in text chats, even more than a human can convince another person that they are human. Also: How to use ChatGPT: A beginner’s guide to the most popular AI chatbotThat’s a breakthrough in the ability of gen AI to produce compelling output in response to a prompt. More
<!–> ZDNET’s key takeaways The OnePlus Watch 2R is available now for $170 (with coupon code ‘LUCKY25’) in Green or Black. It has pretty much everything you look for in a smartwatch: multi-day battery life, fitness tracking features, and a vivid display. However, the device is missing important key fitness metrics, it isn’t as durable […] More
Elyse Betters Picaro / ZDNETThe latest series of trade tariffs announced by the current US administration on “Liberation Day” against trade partners has people worried that their money will be spread even more thinly.Let’s remind ourselves of the tariffs as outlined on the oversized poster board prop used during the event. More
Matthew Miller/ZDNETOnePlus is offering a new promotion for the OnePlus Watch 2 More
<!–> ZDNET’s key takeaways The AirFly Pro 2 allows users to listen wirelessly to traditionally wired-only systems like in-flight entertainment systems. You can use your favorite headphones with the AirFly Pro 2 and enable their special audio features like noise cancellation and spatial audio. If you don’t fly regularly or don’t need a Bluetooth dongle […] More
<!–> ZDNET’s key takeaways Lenovo’s fifth-generation ThinkPad T14s is available now for $2,119. It’s a particularly hardy laptop and boasts solid battery life. However, the Intel chipset delivers puts a cap on high-end performance. more buying choices Lenovo has seven distinct ThinkPad categories on its website, each with a unique style. The T Series, for […] More
Elyse Betters Picaro / ZDNETAs part of my AI coding evaluations, I run a standardized series of four programming tests against each AI. These tests are designed to determine how well a given AI can help you program. This is kind of useful, especially if you’re counting on the AI to help you produce code. The last thing you want is for an AI helper to introduce more bugs into your work output, right?Some time ago, a reader reached out to me and asked why I keep using the same tests. He reasoned that the AIs might succeed if they were given different challenges.Also: Want free AI training from Microsoft? You can sign up for its AI Skills Fest nowThis is a fair question, but my answer is also fair. These are super-simple tests. I’m using PHP and JavaScript, which are not exactly challenging languages, and I’m running some scripting queries through the AIs. By using exactly the same tests, we’re able to compare performance directly. One is a request to write a simple WordPress plugin, one is to rewrite a string function, one asks for help finding a bug I originally had difficulty finding on my own, and the final one uses a few programming tools to get data back from Chrome. But it’s also like teaching someone to drive. If they can’t get out of the driveway, you’re not going to set them loose in a fast car on a crowded highway. To date, only ChatGPT’s GPT-4 (and above) LLM has passed them all. Yes, Perplexity Pro also passed all the tests, but that’s because Perplexity Pro runs the GPT-4 series LLM. Oddly enough, Microsoft Copilot, which also runs ChatGPT’s LLM, failed all the tests. Also: The best AI for coding (and what not to use)Google’s Gemini didn’t do much better. When I tested Bard (the early name for Gemini), it failed most of the tests (twice). Last year, when I ran the $20-per-month Gemini Advanced through my tests, it failed three of the four tests.But now, Google is back with Gemini Pro 2.5. What caught our eyes here at ZDNET was that Gemini Pro 2.5 is available for free, to everyone. No $20 per month surcharge. While Google was clear that the free access was subject to rate limits, I don’t think any of us realized it would throttle us after two prompts, which is what happened to me during testing. It’s possible that Gemini Pro 2.5 is not counting prompt requests for rate limiting but basing its throttling on the scope of the work being requested. My first two prompts asked Gemini Pro 2.5 to write a full WordPress plugin and fix some code, so I may have used up the limits faster than you would if you used it to ask a simple question. Even so, it took me a few days to run these tests. To my considerable surprise, it was very much worth the wait. Test 1: Write a simple WordPress pluginWow. Well, this is certainly a far cry from how Bard failed twice and Gemini Advanced failed back in February 2024. Quite simply, Gemini Pro 2.5 aced this test right out of the gate. Also: I asked ChatGPT to write a WordPress plugin I needed. It did it in less than 5 minutesThe challenge is to write a simple WordPress plugin that provides a simple user interface. It randomizes the input lines and distributes (not removes) duplicates so they’re not next to each other. Last time, Gemini Advanced did not write a back-end dashboard interface but instead required a shortcode that needed to be placed in the body text of a public-facing page. Gemini Advanced did create a basic user interface, but that time clicking the button resulted in no action whatsoever. I gave it a few alternative prompts, and it still failed. But this time, Gemini Pro 2.5 gave me a solid UI, and the code actually ran and did what it was supposed to. More
