Grammarly will soon no longer just recommend ways for you to improve your writing, it’ll do the writing for you.
The writing assistant Grammarly already uses AI in several ways to help it act as a clever tool. Not only can it pick up common grammar and spelling mistakes, but it can also recommend ways to better structure your sentences, and can even tell you the tone your writing portrays (with adjectives like Formal, Confident, Accusatory, and Egocentric).
Come April, Grammarly will be taking its help a step further with the introduction of GrammarlyGo.
Built on OpenAI’s GPT-3 large language models (OpenAI is the team behind ChatGPT), GrammarlyGo will be able to perform a slew of different functions. If you have a document that’s already been written, GrammarlyGo will be able to edit it to portray a different tone or change the length to make your writing clearer or more succinct. Alternatively, if you’re experiencing a writing block its ideation tools will supposedly help unlock your creativity by creating brainstorms and outlines based on prompts you provide.
The press release announcement says it won’t stop at outlines either. GrammarlyGo will be able to compose whole documents for you, and it can even generate replies to emails based on the context of the conversation.
(Image credit: Grammarly)
We haven’t yet had a chance to try GrammarlyGo for ourselves, but we expect it’ll perform similarly to other ChatGPT alternatives we've tested. Specifically, we imagine it’ll show a lot of promise, but its compositions will almost certainly need to be proofread and tweaked by a human – especially while it’s still in beta. Even when given prompts to work with we’ve found that AI writing bots can struggle to generate content that sounds authoritative. Sure, they can produce 400 words about, say, VR headsets, but the writing is often full of chaff and sprinkled with buzzwords rather than feeling like it’s written by someone that understands the topic.
GrammarlyGo’s beta will launch in April (we don’t have an exact date yet) and will be available to all Grammarly Premium, Grammarly Business, and Grammarly for Education subscribers. It’ll also be accessible to people using the free version of Grammarly in the US, UK, Australia, Canada, and New Zealand.
It’s not just writing that OpenAI’s tech is helping to improve. Spotify has launched an AI DJ that can talk to you while mixing your favorite tracks, and Microsoft has incorporated ChatGPT into its search engine to create the impressive Bing Chat tool.
Back in 2011, Apple added its Siri voice assistant to any iPhone running iOS 5 and above, and since then, Siri has made its way onto the whole range of Apple’s products.
Technically speaking, Siri’s original developers should get the credit here – many people still don’t know that it was actually a third-party iOS app for just a few months before Apple acquired it, stopping any plans to bring the software to competitor operating systems like Android and Blackberry.
In response, we saw a veritable uprising from the world’s biggest tech companies to try and compete. Microsoft introduced Cortana in 2013, Amazon Alexa joined the fray in 2014, Google with Google Assistant in 2016, and even more recently we’ve seen newcomers like Bixby and Baidu pop up.
As these applications have battled it out, however, a new player has entered the field with an entirely different skill set that could disrupt the voice assistant space completely; ChatGPT.
Spot the difference
Original comic sourced from @pedro_bilohh on Twitter, edited masterfully by yours truly. (Image credit: Future / Twitter @pedro_bilohh)
So let’s get the differences out of the way first. ChatGPT is an incredibly powerful chatbot with a human-like vocabulary bolstered by near-unfettered access to information. Siri and other voice assistants, alternatively, are programmed to be more binary, with set requests and responses that they can understand.
If you were to ask ChatGPT for assistance in writing or problem-solving, or even some more unique use cases, you’re likely to be surprised and delighted by its capabilities. Powered by the same technology, Bing can also comprehend more challenging questions, even if you ask it about love.
ChatGPT was created by OpenAI, a company which – as its name suggests – allows its technology to be implemented by other organizations, rather than the closed-source proprietary tech that's found in Siri. This means app developers can easily add ChatGPT to all kinds of interesting and exciting apps.
Siri, however, wouldn’t be able to do the same. It’s great for task assistance, especially when boosted by shortcuts, and for quickly navigating tasks hands-free on your phone.
However, it’s frustratingly limited in scope beyond this and struggles to deal with more complex requests even in comparison to Alexa, despite Apple's efforts to enhance it over the years. Plus, I still have a bone to pick with how rubbish its voice recognition can be.
Get with the program, Apple
Microsoft is now stealing the lead in the innovation race with Bing, despite some early teething issues, but Google is hot on its tail. Now, while these are both progressions in the search engine space, it’s only a matter of time before eyes turn to voice assistants.
Siri is used for search, after all – but despite having many years to iterate, search remains one of its most frustrating, clunky features.
Already, keen users are creating ways to embed ChatGPT’s more advanced conversational processing into Siri. It’s far from perfect, and there are some natural, very warranted security and privacy concerns on ChatGPT’s part here, but this eagerness to bolster Siri’s capabilities shows the potential here for Apple to capitalize on.
So, why is Apple dragging its feet?
I wonder if Apple’s fall detection encompasses lagging behind in the innovation race? (Image credit: Apple)
Playing the long game
For me, there’s only one reason a giant like Apple wouldn’t move with pace to recapture the voice assistant market. Like the great tactician Cruella de Vil, Apple may just be biding its time before striking to recapture the voice assistant market.
“You come to realize, you’ve seen her kind of eyes watching you from underneath a rock”
Disney’s 101 Dalmatians
Apple made a slew of AI acquisitions in recent years that we haven’t seen amount to much, and despite a few small rumors indicating something might be coming, the tech giant has been characteristically reserved since the big Bing and Bard blowup.
My take is that Apple was always planning to release something, but I find it hard to believe ChatGPT and Bing didn’t somewhat blindside it. After all, even Google seemed a little pressed to get Bard up and running quickly in response. As a result, Apple was faced with two choices; rush to join the race, or wait and see how the chips fall. Seemingly, it chose the latter.
Now, while Apple wouldn’t stand to lose much by biding its time, it could win big if it comes out with a Siri far more capable than anything else on the market – and if we look at the wider Apple ecosystem and progress elsewhere in the tech space, it seems likely that the company is hoping to kick off with a bang.
Take the smart home space, for example. This year will see Matter, the software standard driving smart home interoperability, begin to really make an impact in people’s homes. If Siri can get the jump on Alexa and Google Assistant, a more conversive and customizable Siri could rocket Apple Home into the lead. Think Disney’s Smart House, but without the murderous vibes.
We’re just a few months out from WWDC, Apple's developer conference in California, which is an event where the company usually showcases its latest software updates (and launches the odd piece of hardware as well). By the time it rolls around, the dust will have settled somewhat on Google and Microsoft’s forays into AI. So long as Amazon doesn’t step in with its own major Alexa overhaul, this could be the perfect opportunity for Apple to sweep in and steal the limelight – potentially, even, with a much more thoroughly thought-out AI.
Time will tell – but one thing is certain: slow and steady may win the race, but not if you never leave the starting line.
Clyde, Discord's cartoon mascot, is getting an AI makeover. On Thursday, the popular chat platform unveiled a wide collection of new AI tools and experiments including an OpenAI-powered chatbot update.
You may not use Discord every day, or at all, but it's become a go-to destination for people looking to have group discussions about topics like gaming, pets, anime, coding, and other nerdy pursuits. According to Discord, 150 million people chat on the service each month.
It's also become a key destination for AI content creation where, according to Discord, 30M users have experienced AI apps on the platform each month.
While most people are now familiar with OpenAI's DALL-E 2 AI image generation tool, Discord users have been creating images with Midjourney's generative AI image tools for just as long. The difference in Discord, though, is the community enhancement that revolves around AI content creation.
As Discord CEO Jason Citron explained it on Tuesday, Discord provides, “AI at your fingertips, together.”
Next week, Discord rolls out the OpenAI-enhanced Clyde as a free public experiment.
Discord Clyde AI (Image credit: Discord)
Like Bing's new Chatbot, Clyde utilizes ChatGPT's large language model to engage in conversation. You invoke Clyde, opt-in to use him on your server and in your group, and can then start asking Clyde questions much in the same way as you'd converse with ChatGPT.
The difference is that because Clyde operates in Discord's chat environment, it has to be aware of group discussion dynamics. Clyde will not just pipe in and interrupt a group chat but a mention of it will allow Clyde to join in. Clyde operates like a real Discord member and can in its responses include GIFs and emojis (ChatGPT in Bing can use emojis, but not GIFs).
Clyde can look up information to settle a dispute or assist with the group project. As with everything else on Discord, admins can easily disable Clyde.
While we've seen a lot of AI chatbot abuse, Discord's Citron prefaced the announcements by reiterating how they work hard to enable all of this “in a safe and trustworthy environment.”
Other AI experiments
Beyond chat, Discord is actively spreading AI capabilities throughout the platform, with many of them launching as limited experiments next week.
To help server moderators, Discord is enhancing its AutoMod tool which already uses preset keywords to actively ban content from group chats with AI that takes proactive moderation a step further.
Discord AutoMod AI (Image credit: Discord)
If, for example, you posted in your group chat that the group does not allow self-promotion or topics outside the core one, which might be sailboats, AutoMod AI could use that post to ban posts skirting those rules. It can even do it if the post is in a different language.
(Image credit: Discord)
Discord chats are often quite active and since no one can spend 24/7 monitoring them, Discord is adding a Conversations Summary Experiment. As the name suggests, it can with a prompt read through previous Discord discussions to surface who was chatting, and the context of all the messages.
If, for example, someone at the end of a long chat asks you if you're coming to “the event on Saturday,” you can ask the AI to summarize the previous chat activity to figure out what event they were talking about – and if you want to attend.
More AI tools
Discord Avatar Remix AI (Image credit: Discord)
We also got a sneak peek at a couple of interesting AI tools, one productivity-focused and the other mostly pure whimsy.
There's the Avatar Remix App, which will let you use a prompt to apply a generative image update to a group member's avatar. Discord showed us how it could easily add a birthday hat to one avatar image and a mustache to the other. The effect is realistic and pretty clever. Obviously, there might be some concern about image abuse here, but at least the tool is only usable on people who already opted into your group.
Discord plans to place the open-source code on GitHub so developers can fork, mix, and extend the Remix code.
(Image credit: Discord)
There's also a powerful-looking new integrated Whiteboard with AI. It'll be the first shared, real-time whiteboard space inside of Discord. The AI integration lets you use simple sketches and text prompts to generate rich, expressive images. Discord claims it can “solve the blank canvas syndrome” many people have with these AI image-generation tools.
To try out any of these AIs, though, you need to be on Discord. That means setting up an account, joining a server, and finding or setting up your own groups. It's not hard but for the uninitiated, Discord can be overwhelming. However, these cool tools might inspire you to give it a try.
ChatGPT might have created the next Wordle in the shape of Sumplete, an engaging online number puzzle game.
The rules of Sumplete are pretty simple. You’re given a grid of numbers – with sizes ranging from a three-by-three grid up to one that’s nine-by-nine – and you have to delete numbers such that the ones remaining in the columns and rows add up to specific totals.
As you can see in the example below, we’ve removed the six and nine from the leftmost column so that the remaining digits (two, two, and three) add up to seven, while for the middle row, we had to delete everything except the central nine to achieve the target of nine.
(Image credit: Future)
As the grids get larger there are more possible combinations of numbers in each row and column that add up to the targets, making it progressively more difficult to find a solution. Those of you who are up for the ultimate challenge will enjoy the nine-by-nine grid, as not only is it the largest, but it introduces negative numbers as well.
If Sumplete isn’t your thing, then you still have Wordle to sink your teeth into, and if you need a hand with today’s puzzle check out our Wordle clues for hints on how to solve it.
How did ChatGPT create Sumplete?
ChatGPT didn’t conjure up Sumplete entirely on its own, and it didn’t create Sumplete on its first attempt at a game either. Instead, it was steered by Substack user Puzzled Penguin.
He initially asked the bot to suggest existing Sudoku-like puzzle games, but he’d played all of them before, so Puzzled Penguin thought the AI bot might be able to create something new. Several iterations later it had Sum Delete – which was later renamed Sumplete – and ChatGPT was even able to create HTML and JavaScript code for the game so that Puzzled Penguin could share the fun.
As for how the idea for Sumplete came about, ChatGPT seems to have combined two games into one. Sudoku is one of its inspirations, but Sumplete shares a few similarities with Picross too; just as in Sumplete, in Picross you use clues about each row and column's contents to identify the correct points in a grid.
Only time will tell if Sumplete has the staying power of Wordle, but even if it doesn’t stick around we expect Sumplete won’t be the last game to be made using ChatGPT’s help.
DuckDuckGo is the latest search engine getting on the generative AI train. It recently launch the beta of its new summarization tool known as DuckAssist, which utilizes “natural language technology” from both OpenAI and AI research company Anthropic.
Though similar to Bing, DuckAssist is not quite like ChatGPT. Instead of utilizing multiple sources to create the summaries, the tool uses primarily just one: Wikipedia. DuckDuckGo specifically chose Wikipedia “because it’s a public resource with a transparent editorial process that cites all the sources used in an article”. The company also points out that since the platform is frequently updated, DuckAssist will always deliver up-to-date information – at most, a few weeks old. Occasionally, the tool will pull from other platforms like Encyclopedia Britannica. However, Wikipedia will be the main one.
Using a single source for information brings with it multiple benefits, according to DuckDuckGo, like being able to generate answers for a vast number of queries quickly. Additionally, only pulling from Wikipedia and its sources reduces the rate of hallucinations – a problem generative AIs have where the tech will just make something up unrelated to the search query.
Work in progress
The way DuckAssist works is pretty simple. All you have to do is ask DuckDuckGo a question, and it’ll immediately write up a summary, complete with the sourced Wikipedia article at the bottom. It’ll even point to the specific section of said article where the original information can be found.
The announcement post gives some suggestions on how to get the most out of DuckAssist. For example, “phrasing your search query as a question [or] adding the word ‘wiki’” increases the chances the summary will appear.
Since the tool is in beta, it’s not perfect. DuckDuckGo admits DuckAssist will not get it right 100 percent of the time. It may omit key information, give the wrong information, cite the wrong source, or all three at once – especially if it’s a particularly complex question. Also, not every query will be given an answer such as asking about recent global events.
Because of these issues, DuckDuckGo is asking users to provide suggestions on DuckAssist and how it can improve the tool. Next to the summaries will be an anonymous feedback link where you can send feedback.
Availability
The tool is currently available on DuckDuckGo’s mobile apps and browser extensions, although not everyone will get to try it out. For those who can, it’s free and totally private. None of the queries will be used to train any AI models nor will OpenAI, Anthropic, or any third-party have access to that information. DuckAssist will roll out to all users within the coming weeks assuming everything goes well with the beta.
It’s worth mentioning this is the first of a series of AI-assisted features that DuckDuckGo is working on. Not much else is known beyond that, but It'll be interesting to see what the developers come up with.
If any of this sounds familiar to you, that’s because Brave launched something very similar on its own web browser. Be sure to check out our coverage of Brave’s Summarizer feature.
In an interview with the Financial Times a few weeks ago, Microsoft’s CEO Satya Nadella dismissed voice assistants, such as Alexa and Siri, as “dumb as a rock”.
This might seem a little rich coming from the CEO of a company that launched (and then abandoned) the unloved Cortana voice assistant, but I actually agree. However, unlike Nadella, I'm not so sure that the new wave of AI chatbots are where the future really lies – or at least not yet.
Sure, they appear to be smarter than the first bunch of voice assistants, including Amazon's Alexa, Apple's Siri and Google's (less charmingly named) Assistant, but that's not saying a lot. I was initially really impressed with these assistants, particularly Alexa, to the extent that I put aside my misgivings about how much information Amazon already collected about me, and filled my home with Echo devices of all shapes and sizes.
That all seems a long time ago now, though. Despite all the promise those voice-activated digital assistants had when they launched, I can't help but feel they’ve turned into little more than hands-free light switches and timers for when I’m cooking. They even made me temporarily forget how to use a real light switch. Seriously.
That’s it. I don’t even use Alexa to play music any more. Partly because none of the Echo devices I have come with remotely decent speakers, and also because Alexa seems to have developed a strange habit where when I ask for a song to be played, it more often than not chooses a random alternative take or live version, rather than the studio version I was after. All very frustrating, especially if you're a Bob Dylan fan.
Even as a light switch, I’ve found it increasingly unreliable. I now often have to repeat myself several times before Alexa understands my request and complies. That’s if I’m lucky. Sometimes it listens, then just does nothing.
It’s become more of an inconvenience and annoyance – the exact opposite of what these virtual assistants were supposed to be. To be fair to Nadella, he told the Financial Times that “Whether it’s Cortana or Alexa or Google Assistant or Siri, all these just don’t work. We had a product that was supposed to be the new front-end to a lot of [information] that didn’t work.”
We’re not alone in getting disillusioned with voice assistants. As the Financial Times reports, Adam Cheyer, co-creator of Apple's Siri, says that “the previous capabilities have just been too awkward… No one knows what they can do or can’t do. They don’t know what they can say or can’t say.”
It also seems like the companies behind the voice assistants are losing interest. Not only did Microsoft unceremoniously dump Cortana after years of trying to get Windows 10 and Windows 11 users to embrace (or at least tolerate) it, Amazon has cut a large number of jobs recently, and there are reports that the teams involved with Alexa and Echo devices have been particularly hard hit.
Two wrongs don’t make a right
It may be easy to suggest that Nadella’s dismissal of voice assistants is down to sour grapes, as Microsoft’s Cortana was the least popular out of the ‘big four’ – which also includes Alexa, Google Assistant, and Siri (sorry, Samsung, but no one likes Bixby either) – but I certainly agree with him. The shine has worn off.
However, it’s increasingly looking like Microsoft thinks that artificial intelligence chatbots, most noticeably ChatGPT, could solve these problems – and it’s here where I’m going to have to disagree, at least for now.
Microsoft is a big investor in ChatGPT and OpenAI, the company behind it, but when it announced it was bringing the power of ChatGPT to its Bing search engine, it managed something rare: it got people excited about Bing.
Suddenly, people were keen to try out a browser which had for so long been neglected in favor of Google. This surge in interest, plus widespread coverage in the press, has deepened Microsoft’s love affair with ChatGPT.
Having an AI bot that can converse with humans in a life-like way, and use huge amounts of stored data in its own libraries and on the internet to answer questions, seems like the natural evolution of voice assistants.
And, one day it might be. However, the technology has so far not lived up to expectations. People using ChatGPT or the version included in Bing have found the chatbot can give incorrect information, and it can also behave strangely, especially if you challenge it when it replies with the wrong answer. A similar issue emerged with Google’s rival Bard AI, which returned an incorrect answer to a question during the launch event. This quickly became quite embarrassing for Microsoft and Google, and it proved to a lot of us that AI bots are not quite ready for the limelight.
Can’t live up to the hype, sometimes unreliable and even a bit frustrating? That certainly sounds familiar, so if Microsoft and other companies don’t want history repeating, they’d do well to think twice before rushing to implement AI bots in voice assistants.
ChatGPT has quickly become one of the most significant tech launches since the original Apple iPhone in 2007. The chatbot is now the fastest-growing consumer app in history, hitting 100 million users in only two months – but it's also a rapidly-changing AI shapeshifter, which can make it confusing and overwhelming.
That's why we've put together this regularly-updated explainer to answer all your burning ChatGPT questions. What exactly can you use it for? What does ChatGPT stand for? And when will it move to the next-gen GPT-4 model? We've answered all of these questions and more below. And no, ChatGPT wasn't willing to comment on all of them either.
In this guide, we'll mainly be covering OpenAI's own ChatGPT model, launched in November 2022. Since then, ChatGPT has sparked an AI arms race, with Microsoft using a form of the chatbot in its new Bing search engine and Microsoft Edge browser. Google has also quickly responded by announcing a chatbot, tentatively described as an “experimental conversational AI service”, called Google Bard.
These will be just the start of the ChatGPT rivals and offshoots, as OpenAI is offering an API (or application programming interface) for developers to build its skills into other programs. In fact, Snapchat has recently announced a chatbot 'called My AI' that runs on the latest version of OpenAI's tech.
For now, though, here are all of the ChatGPT basics explained – along with our thoughts on where the AI chatbot is heading in the near future.
What is ChatGPT?
ChatGPT is an AI chatbot that's built on a family of large language models (LLMs) that are collectively called GPT-3. These models can understand and generate human-like answers to text prompts, because they've been trained on huge amounts of data.
For example, ChatGPT's most recent GPT-3.5 model was trained on 570GB of text data from the internet, which OpenAI says included books, articles, websites, and even social media. Because it's been trained on hundreds of billions of words, ChatGPT can create responses that make it seem like, in its own words, “a friendly and intelligent robot”.
(Image credit: ChatGPT)
This ability to produce human-like, and frequently accurate, responses to a vast range of questions is why ChatGPT became the fastest-growing app of all time, reaching 100 million users in only two months. The fact that it can also generate essays, articles, and poetry has only added to its appeal (and controversy, in areas like education).
But early users have also revealed some of ChatGPT's limitations. OpenAI says that its responses “may be inaccurate, untruthful, and otherwise misleading at times”. OpenAI CEO Sam Altman also admitted in December 2022 that the AI chatbot is “incredibly limited” and that “it's a mistake to be relying on it for anything important right now”. But the world is currently having a ball exploring ChatGPT and, despite the arrival of a paid ChatGPT Plus version, you can still use it for free.
What does ChatGPT stand for?
ChatGPT stands for “Chat Generative Pre-trained Transformer”. Let's take a look at each of those words in turn.
The 'chat' naturally refers to the chatbot front-end that OpenAI has built for its GPT language model. The second and third words show that this model was created using 'generative pre-training', which means it's been trained on huge amounts of text data to predict the next word in a given sequence.
An illustration from Google’s 2017 research paper for the Transformer architecture, which ChatGPT is based on. (Image credit: Google)
Lastly, there's the 'transformer' architecture, the type of neural network ChatGPT is based on. Interestingly, this transformer architecture was actually developed by Google researchers in 2017 and is particularly well-suited to natural language processing tasks, like answering questions or generating text.
ChatGPT was released as a “research preview” on November 30, 2022. A blog post casually introduced the AI chatbot to the world, with OpenAI stating that “we’ve trained a model called ChatGPT which interacts in a conversational way”.
The interface was, as it is now, a simple text box that allowed users to answer follow-up questions. OpenAI said that the dialogue format, which you can now see in the new Bing search engine, allows ChatGPT to “admit its mistakes, challenge incorrect premises, and reject inappropriate requests”.
(Image credit: OpenAI)
ChatGPT is based on a language model from the GPT-3.5 series, which OpenAI says finished its training in early 2022. But OpenAI did also previously release earlier GPT models in limited form – its GPT-2 language model, for example, was announced in February 2019, but the company said it wouldn't release the fully-trained model “due to our concerns about malicious applications of the technology”.
OpenAI also released a larger and more capable model, called GPT-3, in June 2020. But it was the full arrival of ChatGPT in November 2022 that saw the technology burst into the mainstream.
How much does ChatGPT cost?
ChatGPT is still available to use for free, but now also has a paid tier. After growing rumors of a ChatGPT Professional tier, OpenAI said in February that it was introducing a “pilot subscription plan” called ChatGPT Plus in the US. A week later, it made the subscription tier available to the rest of the world.
ChatGPT Plus costs $ 20 p/month (around £17 / AU$ 30) and brings a few benefits over the free tier. It promises to give you full access to ChatGPT even during peak times, which is when you'll otherwise frequently see “ChatGPT is at capacity right now” messages during down times.
(Image credit: OpenAI)
OpenAI says the ChatGPT Plus subscribers also get “faster response times”, which means you should get answers around three times quicker than the free version (although this is no slouch). And the final benefit is “priority access to new features and improvements”, like the experimental 'Turbo' mode that boosts response times even further.
It isn't clear how long OpenAI will keep its free ChatGPT tier, but the current signs are promising. The company says “we love our free users and will continue to offer free access to ChatGPT”. Right now, the subscription is apparently helping to support free access to ChatGPT. Whether that's something that continues long-term is another matter.
How does ChatGPT work?
ChatGPT has been created with one main objective – to predict the next word in a sentence, based on what's typically happened in the gigabytes of text data that it's been trained on.
Once you give ChatGPT a question or prompt, it passes through the AI model and the chatbot produces a response based on the information you've given and how that fits into its vast amount of training data. It's during this training that ChatGPT has learned what word, or sequence of words, typically follows the last one in a given context.
For a long deep dive into this process, we recommend setting aside a few hours to read this blog post from Stephen Wolfram (creator of the Wolfram Alpha search engine), which goes under the bonnet of 'large language models' like ChatGPT to take a peek at their inner workings.
But the short answer? ChatGPT works thanks to a combination of deep learning algorithms, a dash of natural language processing, and a generous dollop of generative pre-training, which all combine to help it produce disarmingly human-like responses to text questions. Even if all it's ultimately been trained to do is fill in the next word, based on its experience of being the world's most voracious reader.
What can you use ChatGPT for?
ChatGPT has been trained on a vast amount of text covering a huge range of subjects, so its possibilities are nearly endless. But in its early days, users have discovered several particularly useful ways to use the AI helper.
Other language-based tasks that ChatGPT enjoys are translations, helping you learn new languages (watch out, Duolingo), generating job descriptions, and creating meal plans. Just tell it the ingredients you have and the number of people you need to serve, and it'll rustle up some impressive ideas.
But ChatGPT is also equally talented at coding and productivity tasks. For the former, its ability to create code from natural speech makes it a powerful ally for both new and experienced coders who either aren't familiar with a particular language or want to troubleshoot existing code. Unfortunately, there is also the potential for it to be misused to create malicious emails and malware.
We're also particularly looking forward to seeing it integrated with some of our favorite cloud software and the best productivity tools. There are several ways that ChatGPT could transform Microsoft Office, and someone has already made a nifty ChatGPT plug-in for Google Slides. Microsoft has also announced that the AI tech will be baked into Skype, where it'll be able to produce meeting summaries or make suggestions based on questions that pop up in your group chat.
Does ChatGPT have an app?
ChatGPT doesn't currently have an official app, but that doesn't mean that you can't use the AI tech on your smartphone. Microsoft released new Bing and Edge apps for Android and iOS that give you access to their new ChatGPT-powered modes – and they even support voice search.
The AI helper has landed on social media, too. Snapchat announced a new ChatGPT sidekick called 'My AI', which is designed to help you with everything from designing dinner recipes to writing haikus. It's based on OpenAI's latest GPT-3.5 model and is an “experimental feature” that's currently restricted to Snapchat Plus subscribers (which costs $ 3.99 / £3.99 / AU$ 5.99 a month).
OpenAI's CEO Sam Altman has confirmed that it's working on a successor to the GPT-3.5 language model used to create ChatGPT, and according to the New York Times this is GPT-4.
Despite the huge number of rumors swirling around GPT-4, there is very little confirmed information describing its potential powers or release date. Some early rumors suggested GPT-4 might even arrive in the first few months of 2023, but more recent quotes from Sam Altman suggest that could be optimistic.
For example, in an interview with StrictlyVC in February the OpenAI CEO said in response to a question about GPT-4 that “in general we are going to release technology much more slowly than people would like”.
He also added that “people are begging to be disappointed and they will be. The hype is just like… We don’t have an actual AGI and that’s sort of what’s expected of us.” That said, rumors from the likes of the New York Times have suggested that Microsoft's new Bing search engine is actually based on a version of GPT-4.
While GPT-4 is unlikely to bring anything as drastic as graphics or visuals to the text-only chatbot, it is expected to improve on the current ChatGPT's already impressive skills in areas like coding. We'll update this article as soon as we hear any more official news on the next-gen ChatGPT technology.
The Artificial Intelligence dream has landed in our everyday lives, and the ethical discussions around AI have ramped up as a consequence, especially concerning how much data these AI services are collecting from users. After all, where there is mass storage of possibly sensitive information, there are cybersecurity and privacy concerns.
Microsoft’s Bing search engine, which is newly equipped with OpenAI’s ChatGPT, and is currently being rolled out, has brought its own set of concerns, as Microsoft hasn’t had the best track record when it comes to respecting its customers’ privacy.
Microsoft has occasionally been challenged about its management and access to user data, although notably less so than its contemporaries like Apple, Google, and Facebook, even though it deals in a great deal of user information – including when it sells targeted ads.
It’s been targeted by certain government regulatory bodies and organizations, such as when France demanded that Microsoft ceases tracking users through Windows 10, and the company responded with a set of comprehensive measures.
Jennifer King, director of consumer privacy at the Center for Internet and Society Stanford Law School, speculated that this is partly due to Microsoft’s longstanding position both in its respective market and long-time relationships with governments afforded to it because of its legacy. It has more experience when dealing with regulators, so might have avoided the same level of scrutiny as its competitors.
An influx of data
Microsoft, as well as other companies, is now finding itself having to react to a mass influx of user chat data due to the popularity of chatbots like ChatGPT. According to the Telegraph, Microsoft has reviewers who analyze user submissions to limit harm and respond to potentially dangerous user inputs by combing through user conversation logs with the chatbot and stepping in to moderate “inappropriate behavior.”
The company claims that it strips submissions of personal information, users' chat texts are only accessible to certain reviewers, and these efforts protect users even when their conversations with the chatbot are under review.
A Microsoft spokesperson elaborated that it employs both automated review efforts (as there is a great deal of data to comb through) and manual reviewers. It goes on to state that this is the standard for search engines, and is also included in Microsoft’s privacy statement.
The spokesperson is at pains to reassure those concerned that Microsoft employs industry-standard user privacy measures such as “pseudonymization, encryption at rest, secured and approved data access management, and data retention procedures.”
Additionally, the reviewers can only view user data on the basis of “a verified business need only, and not any third parties.” Microsoft has since updated its privacy statement to summarize and clarify the above – user information is being collected and human employees at Microsoft may be able to see it.
Under the spotlight
Microsoft isn’t the only company under scrutiny over how it collects and handles user data when it comes to AI chatbots. OpenAI, the company that created ChatGPT, also disclosed that it reviews user conversations.
These concerns are multiplied when considering the usage of ChatGPT and ChatGPT-equipped bots by those working at companies with their own sensitive and confidential information, many of which have warned employees not to submit confidential company information into these chatbots. Some companies, such as JP Morgan and Amazon, have restricted or banned their use at work altogether.
Personal user data has been, and continues to be, a key issue in tech in general. Misuse of data, or even malicious use of data, can have dire consequences both for individual people and for organizations. With every introduction of a new technology, these risks are increased – but so is the potential reward.
Tech companies would do well to pay extra attention to make sure our personal data is as secure as possible – or lose the trust of their customers and potentially kill off their fledgling AI ambitions.
Director of sales engineering APJ at BlackBerry Cybersecurity, Jonathan Jackson, explains why artificial intelligence poses such a threat in the cyber security space
Hopping on the AI train, Brave is incorporating its own AI-powered search function to its web browser called Summarizer – similar to what Microsoft recently did to Bing.
The new feature “provides concise and to-the-point answers at the top of Brave Search results”. For example, if you want to learn about the chemical spill in East Palestine, Ohio, Summarizer will create a one paragraph summary of the event alongside some sources for you to read. Unlike Microsoft which only uses ChatGPT for Bing's chatbot, Summarizer uses three in-house large language models, LLMs for short, based on retrained versions of the BART and DeBERTa AI models to create the search-results snippets.
Retraining AI
To simplify the technology behind them, BART and DeBERTa are generative writing AIs like ChatGPT that have been specially trained to take into account word positioning as well as context so the text output reads well. What Brave did is take those models and retrain them using its own search result data to develop Summarizer.
Summarizer’s training regiment is a three-step process, according to the announcement. First, Brave taught the LLMs to prioritize answering the question being asked. Then, the company utilized “zero-shot classifiers” to categorize results so the given information is relevant. The final step helps the models rewrite the snippet so it’s more coherent. The result is an accurate answer written succinctly with multiple sources attached.
Be aware the feature is still in the early stages. Brave states Summarizer only utilizes about 17 percent of search queries to formulate an answer, but there are plans to scale that number even higher for better paragraphs. Its accuracy needs some work, too. The company admits Summarizer may produce what it calls “hallucinations” which are unrelated snippets mixed in with results. Plus there's the possibility of the feature throwing in some “false or offensive text” into an answer.
Availability
Summarizer is currently available to all Brave Search users on desktop and mobile with the exception of the Brave Search Goggles. It’s disabled there. You can turn it off anytime you want by going into the browser’s settings menu. The company is also asking users to give some feedback on how it can improve the tool.
We tried out Summarizer ourselves, and as cool as it is, it does need some work. Not all search results will give you a snippet as it depends on what you ask, as well as which news topics are making the rounds. The East Palestine, Ohio chemical spill, for example, is currently a hot button issue so you get Summarizer working just fine there. However when we asked about the recent cold snap in Los Angeles and what’s going on with certain video game developers, we either got no summary or outdated information. But the latter did come with sources so it was at least accurate. Still better than having ChatGPT throw a temper tantrum or lie to your face.
Be sure to check out TechRadar’s list of the best AI writer for 2023 if you’re interested in learning what AI creativity can do for you.
1 844.742.7448 
Email Us
Remote Support 
Contact
