1 844.742.7448

Email Us

Remote Support

Contact

Posts tagged "Wrong"

  1. Home
  2. Posts tagged Wrong

Google explains how Gemini’s AI image generation went wrong, and how it’ll fix it
  • No Comments

A few weeks ago Google launched a new image generation tool for Gemini (the suite of AI tools formerly known as Bard and Duet) which allowed users to generate all sorts of images from simple text prompts. Unfortunately, Google’s AI tool repeatedly missed the mark and generated inaccurate and even offensive images that led a lot of us to wonder – how did the bot get things so wrong? Well, the company has finally released a statement explaining what went wrong, and how it plans to fix Gemini. 

The official blog post addressing the issue states that when designing the text-to-image feature for Gemini, the team behind Gemini wanted to “ensure it doesn’t fall into some of the traps we’ve seen in the past with image generation technology — such as creating violent or sexually explicit images, or depictions of real people.” The post further explains that users probably don’t want to keep seeing people of just one ethnicity or other prominent characteristic. 

So, to offer a pretty basic explanation for what’s been going on: Gemini has been throwing up images of people of color when prompted to generate images of white historical figures, giving users ‘diverse Nazis’, or simply ignoring the part of your prompt where you’ve specified exactly what you’re looking for. While Gemini’s image capabilities are currently on hold, when you could access the feature you’d specify exactly who you’re trying to generate – Google uses the example “a white veterinarian with a dog” – and Gemini would seemingly ignore the first half of that prompt and generate veterinarians of all races except the one you asked for. 

Google went on to explain that this was the outcome of two crucial failings – firstly, Gemini was showing a range of different people without considering a range not to show. Alongside that, in trying to make a more conscious, less biased generative AI, Google admits the “model became way more cautious than we intended and refused to answer certain prompts entirely – wrongly interpreting some very anodyne prompts as sensitive.”

So, what's next?

At the time of writing, the ability to generate images of people on Gemini has been paused while the Gemini team works to fix the inaccuracies and carry out further testing. The blog post notes that AI ‘hallucinations’ are nothing new when it comes to complex deep learning models – even Bard and ChatGPT had some questionable tantrums as the creators of those bots worked out the kinks. 

The post ends with a promise from Google to keep working on Gemini’s AI-powered people generation until everything is sorted, with the note that while the team can’t promise it won’t ever generate “embarrassing, inaccurate or offensive results”, action is being taken to make sure it happens as little as possible. 

All in all, this whole episode puts into perspective that AI is only as smart as we make it. Our editor-in-chief Lance Ulanoff succinctly noted that “When an AI doesn't know history, you can't blame the AI.” With how quickly artificial intelligence has swooped in and crammed itself into various facets of our daily lives – whether we want it or not – it’s easy to forget that the public proliferation of AI started just 18 months ago. As impressive as the tools currently available to us are, we’re ultimately still in the early days of artificial intelligence. 

We can’t rain on Google Gemini’s parade just because the mistakes were more visually striking than say, ChatGPT’s recent gibberish-filled meltdown. Google’s temporary pause and reworking will ultimately lead to a better product, and sooner or later we’ll see the tool as it was meant to be. 

You might also like…

TechRadar – All the latest technology news

Read More

Canva thinks you’re doing data visualization wrong
  • No Comments

Canva thinks businesses are doing data wrong. And it’s rolling out a tool to fix that. 

The company has announced the launch of interactive data visualizations built right into the graphic design software. These are engaging data maps, charts, graphs that make things seem just that little bit clearer. Most users will have toyed with them on news sites like the BBC. Now, they can add them to their own photos, PDFs, and presentations.

The move follows the acquisition of UK data-viz platform Flourish, as the Aussie firm ramps up its Europe presence.

Data ≠ dull

Data is tricky to present – especially to an audience who may be unfamiliar or unengaged with the topic. An Excel-generated bar graph, faded and static, will struggle in a content-rich world. The death yawn of a thousand sales pitches and PowerPoint slides. The last sigh of a think-piece.

And while the company didn’t put it quite like that, it’s this thinking – that data needs to be like everything else: visual – driving the roll-out. So, there’s no excuse for making data boring. 

Read more

> Canva reveals 'magic' new AI apps for marketers

> These are the best desktop publishing software choices around

> Canva unveils new workplace suite to boost your productivity 

Users will already find standard charts and graphs on the platform – we’ve always found them somewhat basic and uninspired. But from today, users can embed Flourish visualizations into their designs. Users get access to native hierarchical treemap and packed circle charts straight from the app, with Canva promising “animated charts, zoom-able maps, explorable diagrams, and more.”

That Flourish is deepening integration into Canva will come as no surprise. It’s not the first by any measure, with the firm counting stock photo sites Pexels and Pixabay amongst its relatively recent acquisitions. Meanwhile, it’s been quietly building out the platform with a free PDF editor, website builder, and AI photo tools. 

Readers may spot a common thread among the acquisitions: they’re all European businesses. Then came the opening of its first European campus in London. It seems the continent is where Canva sees its growth potential, with the company calling it home to some of its “fastest growing and most densely populated markets.” Just don’t tell Adobe Express about Canva’s continental plans.  

TechRadar – All the latest technology news

Read More

Duolingo’s new GPT-4 AI will happily explain why your Spanish is wrong
  • No Comments

Duolingo is launching a new virtual tutor that aims to replicate real-world scenarios to help students learn better. And it’s all powered by the recently released GPT-4 AI model.

Making its home in the new Duolingo Max subscription tier, the tutor consists of two features: Explain My Answer and Roleplay. The former, as its name suggests, gives users the opportunity to, if they’re confused by something in DuoLingo's initial response, ask the chatbot Duo to give a detailed explanation of why their answer was right or wrong. In an example video, the AI explains why select Spanish verbs must be conjugated a certain way given the context of the sentence. 

Duo, however, is not universally available on all language exercises, just certain ones. On those few, a Explain My Answer button will appear at the bottom of the screen after you attempt an exercise.

Roleplay, on the other hand, allows users to engage in a realistic conversation with the AI so they can practice their language skills. According to the post, no two chats will be exactly the same. In one instance, you could be talking to “waiter” as you order coffee at a French café or discussing vacation plans in Spanish with a “friend.” And at the end of every Roleplay, Duo will give you some feedback based “on the accuracy and complexity of [your] responses, as well as tips for future conversations.” 

Limited release

Do be aware that the GPT-4 AI behind Duo is not perfect. For the new release, research laboratory OpenAI took the time to improve GPT-3’s chat abilities so it can produce more natural-sounding text, similar to how people normally speak – at least in English. GPT-4 can create language guides such as utilizing English mnemonics for Spanish words. However, as seen on Twitter, those mnemonic guides can be pretty hilarious and not always in a good way. Duolingo admits its virtual tutor will make some mistakes. As such, the company is asking users to give the AI some feedback which you can do by selecting either the “thumbs-up” or “thumbs-down” emoji at the end of every Explain My Answer session.

Currently, Duolingo Max is seeing a limited roll-out. The AI will only be available in either Spanish or French for English speakers on iOS, but there are plans to “expand to more courses, language interfaces, and platforms in the coming months”, according to a company representative.

To subscribe to the tier, you have two methods. You can either pay $ 29.99 for Duolingo Max or $ 167.99 for the whole year. Breaking everything down, the yearly cost comes down to $ 13.99 a month. Additionally, you also get every feature under Super Duolingo which includes “unlimited hearts [for lessons], no ads, and [a] personalized review through the Practice Hub.”

While we have you, be sure to check out TechRadar’s list of the best AI writers for 2023 if you need content done fast. 

TechRadar – All the latest technology news

Read More

Microsoft slowly but surely admits it was wrong about the Windows 11 taskbar
  • No Comments

Windows 11 just got its first preview build in the new Canary channel – the earliest test versions of the OS going forward, possibly representing the foundations of Windows 12 – and while nothing much of note was introduced on the face of it, a hidden gem has been unearthed.

Twitter-based leaker @PhantomOfEarth pointed out @XenoPanther’s discovery of a bunch of strings tucked away under the hood in build 25314 referring to taskbar grouping.

See more

What does this mean? Well, it’s a tantalizing hint that as previously rumored, Microsoft is going to bring back the ‘never combine’ option when it comes to grouping apps on the taskbar.

In other words, rather than having multiple instances of the same app automatically grouped together on the taskbar (stacked vertically), you can have each of them separated into individual entries (horizontally) on the bar.

Elsewhere in build 25314, there are some minor tweaks, the most significant of those being a change to File Explorer, namely the addition of Access Keys. These are single keystroke shortcuts labeled by a single letter in the context menu of File Explorer – simply hit the relevant key to swiftly execute the command in question.

Furthermore, those using Azure Active Directory will now see recommendations for files they might find useful or relevant at the top of File Explorer Home.

Analysis: It seems like Microsoft is finally listening on the taskbar

When cooking up Windows 11, Microsoft made some mystifying decisions with the interface, leaving out some core bits of functionality seen in Windows 10, most notably with the taskbar. The ability to never combine (stack up) running instances of the same app was one of those features that got dropped.

To see a glimmer of hope that it might be inbound for the future, then, is certainly welcome. Although we still question exactly why it has taken so long for Microsoft to look at implementing this. And we must remember, this is only tinkering in the background in early testing for now – eventually it’s possible nothing could come of it, though we’re trying not to entertain that possibility, frankly. The lack of this feature is a deal-breaker for us, personally, in Windows 11.

As for the other notable taskbar omissions Microsoft made with Windows 11, drag-and-drop support was returned to the bar not so long ago. And in the future, we may also see the resurrection of the ability to move the taskbar from the bottom of the screen to the sides or top. (Currently, it’s locked down at the foot of the screen for Windows 11 users, whereas those on Windows 10 can move it around, of course).

So, it seems that Microsoft is slowly rethinking and reversing course on its taskbar philosophy with Windows 11, and frankly, it’s about time. Especially given all the feedback and voices shouting about these bits of functionality being stripped away for no good reason – not that we can think of, anyway. And don’t give us any excuses about streamlining or simplifying the UI, these can be options in Settings that no one who’s bothered about this sort of thing ever has to look at.

In short, Microsoft, please keep going along this path of reversal, because, you know, we’d like to get back to a Windows 10 level of functionality with the taskbar, if that’s okay?

TechRadar – All the latest technology news

Read More

Twitter’s latest meltdown proves Elon Musk is still doing it wrong
  • No Comments

Twitter thrives on shares, not just within the social media platform but from partner links all over the Internet. Except on Monday, most of those links stopped working.

For approximately an hour, anyone trying to share recently published articles on Twitter was met with an error message clearly intended for developers:

Twitter API bug

(Image credit: Future)

It was almost as if Twitter was informing publishers that they didn't pay their water bill and, as such, couldn't publish links on the social network.

What went wrong?

We didn't have to wait too long for Twitter CEO Elon Musk to explain. In response to a tweet from former Netscape founder and well-known venture capitalist Marc Andreessen pointing out how four of the five top Twitter trends were about Twitter, Musk tweeted, “A small API change had massive ramifications. The code stack is extremely brittle for no good reason. Will ultimately need a complete rewrite.”

See more

This seemingly clear-headed tweet though should be cause for alarm. Musk claims the code stack (basically a massive stack of programs that all work together to create the Twitter whole) is brittle and needs a rewrite. What he fails to mention is that among the thousands of Twitter employees he laid off since November, a good number of them were engineers and, it's safe to assume, some were in what's known as QA or quality assurance.

Typically if you plan on making any kind of code change to a website, online service, or app, QA tests it on an offline copy of the platform. In this way, they ensure that the updates, no matter how small, won't adversely impact the live environment.

The concept is known as “production,” the live site or service, versus “staging,” an environment that's identical to live but can not be seen or touched by users. You run your new code or feature through staging, a group of QA testers applies a set of known scenarios (maybe they throw in an edge case or two) and as long as there are no red flags, the update gets pushed from Staging to Production. 

Twitter, which has seen its overall reliability drop (from going offline to having features appear and disappear unexpectedly) since Musk took over, may be getting its updates in a different way.

Musk likes to test features on production (the live site). As a result, he keeps running into unintended consequences.

There is some disagreement on whether or not there is a Twitter QA team.

Some argue one exists but Musk grows impatient and then pushes untested code live.

Others insist that Elon Musk arrived at Twitter and discovered that Twitter had no QA team and it was long in the practice of pushing untested code live. That though seems highly unlikely. 

I asked Musk directly on Twitter if the API update was tested on staging before being pushed live and will update this post if he responds.

Never assume

The assumption he made here, that a small API change would have little impact on the site was a poor one. And, yet, Musk still doesn't understand that he's doing it wrong.

Testing features of any kind on a live version of a complex platform like Twitter will inevitably result in bugs and crashes.

Will rewriting the code stack solve all this? Maybe, but very few platforms stay as clean as they were on launch and even if the rewrite is robust and perfect, frequent updates and fresh features will test that stability.

As long as Musk refuses to fully test what he launches before he launches it, there is no scenario in which Twitter escapes regular downtime.

This is a simple fix, Elon, make QA an inescapable part of the development pipeline and save yourself and us a lot of headaches. Or keep doing it your way because that's working out so, so well.

TechRadar – All the latest technology news

Read More

WhatsApp will soon help avoid the embarrassment of sending the wrong voice message
  • No Comments

WhatsApp’s voice messaging feature will soon get some nifty updates to help compose, send, and listen to those convenient audio messages.

The Meta-owned app is the main mode of communication for over two billion people every month. It’s free, highly accessible, and end-to-end encrypted, making it an important app for users around the globe to connect with family and friends. These updates could enhance the already pretty good voice messaging feature of the app by helping avoid miscommunications in audio messages and helping listeners speed through long-winded conversations. 

Included features are:

WhatsApp didn’t provide a release date in the announcement or information about which platforms it will arrive on and in what order, but you can expect the features to roll out over the next few weeks. 

WhatsApp Voice Message Update

(Image credit: WhatsApp)

 Analysis: WhatsApp stays on top for a reason 

 The updates that Meta steadily brings to WhatsApp aren’t anything groundbreaking, and that’s by design. There’s a reason that the app continues to be the most popular global messaging app out there. 

Small features brought about in incremental updates maintain the app’s ease of use by not getting in the way of how the app’s two billion monthly active users already interact with it.

With these updates, it looks like the majority of the interface remains the same, and the new draft previews will help users avoid sending messages that weren’t ready yet. It’s the little things that count the most.

TechRadar – All the latest technology news

Read More

WhatsApp will soon help avoid the embarrassment of sending the wrong voice message
  • No Comments

WhatsApp’s voice messaging feature will soon get some nifty updates to help compose, send, and listen to those convenient audio messages.

The Meta-owned app is the main mode of communication for over two billion people every month. It’s free, highly accessible, and end-to-end encrypted, making it an important app for users around the globe to connect with family and friends. These updates could enhance the already pretty good voice messaging feature of the app by helping avoid miscommunications in audio messages and helping listeners speed through long-winded conversations. 

Included features are:

WhatsApp didn’t provide a release date in the announcement or information about which platforms it will arrive on and in what order, but you can expect the features to roll out over the next few weeks. 

WhatsApp Voice Message Update

(Image credit: WhatsApp)

 Analysis: WhatsApp stays on top for a reason 

 The updates that Meta steadily brings to WhatsApp aren’t anything groundbreaking, and that’s by design. There’s a reason that the app continues to be the most popular global messaging app out there. 

Small features brought about in incremental updates maintain the app’s ease of use by not getting in the way of how the app’s two billion monthly active users already interact with it.

With these updates, it looks like the majority of the interface remains the same, and the new draft previews will help users avoid sending messages that weren’t ready yet. It’s the little things that count the most.

TechRadar – All the latest technology news

Read More