Google Lens will soon search for words and images combined

Google is updating its visual search tool Google Lens with new AI-powered language features. The update will let users further narrow searches using text. So, for example, if you snap a photo of a paisley shirt in order to find similar items online using Google Lens, you can add the command “socks with this pattern” to specify the garments you’re looking for.

Additionally, Google is launching a new “Lens mode” option in its iOS Google app, allowing users to search using any image that appears while searching the web. This will be available “soon,” but it’ll be limited to the US. Google is also launching Google Lens on desktop within the Chrome browser, letting users select any image or video when browsing the web to find visual search results without leaving their tab. This will be available globally “soon.”

These updates are part of Google’s latest push to improve its search tools using AI language understanding. The updates to Lens are powered by a machine learning model that the company unveiled at I/O earlier this year named MUM. In addition to these new features, Google is also introducing new AI-powered tools to its web and mobile searches.

Using the updated Google Lens to identify a bike’s derailleur.
Image: Google

The changes to Google Lens show the company hasn’t lost interest in this feature, which has always shown promise but seemed to appeal more as a novelty. Machine learning techniques have made object and image recognition features relatively easy to launch at a basic level, but, as today’s updates show, they require a little finesse on the part of the users to be properly functional. Enthusiasm may be picking up, though — Snap recently upgraded its own Scan feature, which functions pretty much identically to Google Lens.

Google wants these Lens updates to turn its world-scanning AI into a more useful tool. It gives the example of someone trying to fix their bike but not knowing what the mechanism on the rear wheel is called. They snap a picture with Lens, add the search text “how to fix this,” and Google pops up with the results that identify the mechanism as a “derailleur.”

As ever with these demos, the examples Google is offering seem simple and helpful. But we’ll have to try out the updated Lens for ourselves to see if AI language understanding is really making visual search more than just a parlor trick.

Repost: Original Source and Author Link


OpenAI’s text-generating system GPT-3 is now spewing out 4.5 billion words a day

One of the biggest trends in machine learning right now is text generation. AI systems learn by absorbing billions of words scraped from the internet and generate text in response to a variety of prompts. It sounds simple, but these machines can be put to a wide array of tasks — from creating fiction, to writing bad code, to letting you chat with historical figures.

The best-known AI text-generator is OpenAI’s GPT-3, which the company recently announced is now being used in more than 300 different apps, by “tens of thousands” of developers, and producing 4.5 billion words per day. That’s a lot of robot verbiage. This may be an arbitrary milestone for OpenAI to celebrate, but it’s also a useful indicator of the growing scale, impact, and commercial potential of AI text generation.

OpenAI started life as a nonprofit, but for the last few years, it has been trying to make money with GPT-3 as its first salable product. The company has an exclusivity deal with Microsoft which gives the tech giant unique access to the program’s underlying code, but any firm can apply for access to GPT-3’s general API and build services on top of it.

As OpenAI is keen to advertise, hundreds of companies are now doing exactly this. One startup named Viable is using GPT-3 to analyze customer feedback, identifying “themes, emotions, and sentiment from surveys, help desk tickets, live chat logs, reviews, and more”; Fable Studio is using the program to create dialogue for VR experiences; and Algolia is using it to improve its web search products which it, in turn, sells on to other customers.

All this is good news for OpenAI (and Microsoft, whose Azure cloud computing platform powers OpenAI’s tech), but not everyone in startup-land is keen. Many analysts have noted the folly of building a company on technology you don’t actually own. Using GPT-3 to create a startup is ludicrously simple, but it’ll be ludicrously simple for your competitors, too. And though there are ways to differentiate your GPT startup through branding and UI, no firm stands to gain as much as from the use of the technology as OpenAI itself.

Another worry about the rise of text-generating systems relates to issues of output quality. Like many algorithms, text generators have the capacity to absorb and amplify harmful biases. They’re also often astoundingly dumb. In tests of a medical chatbot built using GPT-3, the model responded to a “suicidal” patient by encouraging them to kill themselves. These problems aren’t insurmountable, but they’re certainly worth flagging in a world where algorithms are already creating mistaken arrests, unfair school grades, and biased medical bills.

As OpenAI’s latest milestone suggest, though, GPT-3 is only going to keep on talking, and we need to be ready for a world filled with robot-generated chatter.

Repost: Original Source and Author Link

Tech News

Microsoft teases Word’s next default font — so we got a designer to weigh in

Big, terrifying changes are afoot: there’s going to be a new default font in Microsoft Word. Please, don’t panic. You can riot, sure, but no panicking.

This decision was announced on Microsoft’s blog. In the piece, the company explains that it has commissioned five different fonts that could potentially replace the current default, Calibri.

The piece itself provides a balanced view of all these different options, going into an admirable amount of depth about why they may be suitable to become Microsoft Word’s next default font.

Unfortunately though, we didn’t see this news on the blog. Instead, we saw it on that bastion of rational discussion, Twitter. Have a look:

Is a social network famous for being one of the most toxic environments on the internet the best place to engage in dialogue about a default font? Yes. Definitely. Why bother even asking that?

Anyway, we decided to ask TNW’s VP of Design, Alexander Griffioen, to wade in with his opinions.

But in order to keep the spirit of Microsoft’s social media post alive, we only showed him the image below and asked him to provide only Tweet-length comments. We’re nothing but precise and fair.

Tech News

CharaChorder Lite lets you type whole words like you’re playing a piano

If typing one key at a time on a QWERTY keyboard isn’t fast enough for you, the new CharaChorder Lite keyboard offers an interesting alternative. The product looks just like an ordinary QWERTY keyboard but allows users to type entire words at a time using ‘chords’ as if they were playing the piano. The result, the team claims, is a typing speed of up to 250 words per minute.

It’s possible to type pretty fast with a QWERTY keyboard by hitting one key at a time — but you’ll still struggle to get anywhere near the speed at which you think and speak. CharaChorder Lite aims to change that by enabling users to press a combination of keys at the same time to produce entire words, with the keyboard filling in the blanks.

The keyboard supports typing one letter at a time for those moments when you can’t remember the ‘chord’ for the word you want, alternating between pecking and mashing to ultimately increase the rate at which you produce text. The keyboard would no doubt require some time to get used to, but the idea is fascinating regardless.

The CharaChorder Lite will be made with hot-swap smt connectors for use with standard keyboard switches; it’ll ship with mechanical switches and include an LED-powered backlight. The keyboard will feature ordinary USB-C connectivity and, the team behind the product claims, it will be compatible with any computing device you have, including your desktop, phone, and tablet.

In addition to the chords that are preset for the device, CharaChorder Lite users will be able to create their own chord shortcuts for use with the keyboard. The team claims that learning the 100 most common keyboard chords and using regular typing for the rest of the words will still enable you to mash out around half of everything you ordinarily type.

The CharaChorder Lite is being funded on Kickstarter where it has exceeded its funding goal. Backers who pledge at least $99 are offered the ‘early bird’ units at what amounts to a $200 discount, according to the team. Shipping to early bird backers is estimated to start this December.

Repost: Original Source and Author Link