Generative Ai hits educaton,ushering a sea of change for schools.
September 14, 2023
SMART SPEAKER CO PLAY
September 14, 2023
Generative Ai hits educaton,ushering a sea of change for schools.
September 14, 2023
SMART SPEAKER CO PLAY
September 14, 2023

Generative artificial intelligence (AI) has come leaps and bounds since it was first unveiled just a year ago. The technology has shifted from something seemingly quite abstract to become a technology that’s embedded in a huge variety of readily accessible tools and services.

From image generation – like in Midjourney – to programming – such as GitHub CoPilot – professionals in so many industries can access generative AI tools to assist them in their roles.

The use cases for generative AI are growing. With the field maturing as each day passes, there are also a variety of tools professionals can incorporate into their workflows. We’ve rounded up 11 amazing AI tools you can incorporate into your workflow today.

If any company should be worried about the emergence of generative AI art it’s Adobe, which has enjoyed a stranglehold on computer-generated art for the past couple of decades.

Adobe has been dabbling with generative art with Firefly, a beta web service that allows users to create images from text prompts, fill in text with specified textures, or recolor vector art images. But its big breakthrough arrived with the inclusion of Firefly in the Photoshop beta. It’s nothing short of stunning.

Taking a sample image of a runner with her dog. I extended the canvas, drew a selection box around the empty space, and told Photoshop to fill it with a “pond with waterlilies” and within seconds it came back with the image we see above. The pond edge looks entirely natural, the lighting is right, and it’s even made a decent stab of capturing the runner and dog’s reflection in the water. The reflection’s not quite right, but it’s nothing that a short bit of work with Photoshop’s regular cloning tools wouldn’t easily fix.

Jobs that would previously have taken a fair while and no little amount of skill to pull off convincingly are now performed in seconds. It’s the kind of feature that justifies that huge monthly fee to Adobe.

When you first start using ChatGPT Plus (at a cost of $20 per month), you wonder how on Earth OpenAI justifies such a fee. At the time of writing, even paying customers are rate-limited to 25 messages every three hours using GPT-4. Even though it can now access the web to extend its knowledge beyond the September 2021 cut-off date of the training data, so can the GPT-4-powered AI in Bing – and that’s entirely free.

Then you find the plugins and you suddenly begin to see why we’ve not even started to scratch the surface of what AI can do yet. Enable the OpenTable plugin and tell ChatGPT you want to book a table for four at an Italian restaurant in Brighton on Saturday night and it will go off and present you with a selection of relevant options. When you click through to the chosen restaurants, all the details are pre-filled – you just click to book.

Need something a bit more businesslike? The Code Interpreter plugin can take data trapped in Excel spreadsheets and convert it into, among many other things, 3D scatter plots, with your chosen color scheme and particular data strands highlighted. You just tell the AI the output you want from the data, and it hands you the graph in HTML5, ready to upload to your website. It can even provide you with a written analysis of the data. The plugins are still in their infancy, but even from the hundred or so that are already in the store at the time of writing, you can see the enormous potential here.

Working in the tech industry involves wading through pages of lengthy, poorly written online documentation. Webpilot is an AI assistant that can cut through the blather and extract the key information.

It’s a simple Chrome extension, although you’ll need an OpenAI API key to make the AI work. Once you’ve plugged in your API key, you can visit any web page, click the Webpilot icon at the top of the screen, and ask it to, say, summarise the page in five bullet points. It’s fast, accurate, and might save you an awful lot of time scrolling through page after page.

Webpilot is more than a summarizer. It can help you rewrite text, although AI-generated text is rather flat and has ticks of its own, such as starting sentences with “therefore” and “however”. It can also generate opposing viewpoints, letting you highlight a passage of text and then generate a counter view.

Ever wondered what your voice would sound like if you were a warm, kind, mid-sixties woman from Scotland called Jennifer?

Altered Studio uses AI to transform your speech into a completely different voice. Say, for example, you wanted a 20-something woman to voice your new YouTube ad – but didn’t want the expense and hassle of booking a voice actor. You can record the audio yourself, upload it to Altered Studio, and pick from a huge bank of AI-generated voice characters.

The voices are unbelievably convincing. You’d struggle to tell them apart from real voice actors. Visit the site to listen to various samples, or take up the free trial and experiment with a limited selection available to subscribers.

There’s a huge range of potential uses for Altered Studio. It may provide a second or third podcast host for a podcast or can introduce speakers at a conference. In our tests, it worked best when you provided a voice recording – it also offers text-to-speech, but that emerged with a robotic speech pattern that clearly marked it out as AI.

With dozens of text-to-image services out there, it’s impossible to say which is best. But the one service I consistently turn to for AI-generated art is Midjourney, which continues to deliver consistently high-quality results across a wide variety of styles and genres.

The key to its success is the rather byzantine way you access the service. There’s no direct web interface; instead, you enter your prompts via the Discord chat service and wait for your results to appear in the chat thread. It’s a weird way of operating, but it has two distinct advantages. First, it works seamlessly on desktop and mobile apps, on pretty much any platform. Secondly, it encourages you to browse and learn from what others are doing, with everyone’s results available to view in the various chat threads. Prompts appear alongside images, allowing you to take inspiration.

Midjourney also offers a degree of control that’s rare in generative art circles. The key element here is the parameters that, for example, let you change the aspect ratio by adding “–ar 16:9” to your prompt, or the degree of “chaos” in the image. Midjourney is also a fantastic mimic. Tell it to create an image “in the style of” Van Gogh, or Matt Groening, or Robert Capa, and that’s what you’ll get.

It isn’t free – you’ll need to pay $10 a month to access the service, and even that imposes (generous) usage limits. But Midjourney’s consistently jaw-dropping results merit reward.

I’m a long way from a professional coder. I know my way around HTML/CSS files, have dabbled with JavaScript and Python, but that’s as far as it goes. In some ways, that makes me the perfect candidate to test drive GitHub Copilot.

Copilot is autocomplete for code. It’s a code-editor extension that sits quietly in the background, until you start entering or tweaking code on a page, after which you’ll likely see suggested edits appear in grey.

For coding novices, it’s particularly useful when used in conjunction with AI-written code. For example, I asked Bing AI to help me write a basic VAT calculator – you enter a price, hit a button, and it shows you what the price is exc VAT. I asked it to deliver HTML, CSS, and JavaScript files and it did so perfectly.

So now I’ve got my lovely VAT calculator I obviously want to create an About page for my website, so people know where to hire this coding genius. I create an about.html and start to enter at the top of the page, and Copilot takes over. It suggests a with About in the title tags and automatically creates a link to the correct stylesheet. It even starts suggesting copy for the page, as you can see in the screenshots, writing: “This is a simple VAT calculator. It calculates the price without VAT from the price with VAT.” It’s not going to win any prizes for elegance, but the AI has written an explanation of what the site does without any prompting.

This is a basic example, but it’s capable of far more. For example, you can write in comments exactly what to do, and CoPilot will suggest the necessary code. Instantly. Right there in Visual Studio Code or other popular editors. Often with multiple suggestions, in case you don’t like the first solution it offers.

It’s no replacement for coding knowledge. You’ll still need to know how the plumbing works and it will get things wrong from time to time. But GitHub Copilot is almost guaranteed to make your coding jobs monumentally easier for only $10 per month.

So now I’ve got my lovely VAT calculator I obviously want to create an About page for my website, so people know where to hire this coding genius. I create an about.html and start to enter at the top of the page, and Copilot takes over. It suggests a with About in the title tags and automatically creates a link to the correct stylesheet. It even starts suggesting copy for the page, as you can see in the screenshots, writing: “This is a simple VAT calculator. It calculates the price without VAT from the price with VAT.” It’s not going to win any prizes for elegance, but the AI has written an explanation of what the site does without any prompting.

This is a basic example, but it’s capable of far more. For example, you can write in comments exactly what to do, and CoPilot will suggest the necessary code. Instantly. Right there in Visual Studio Code or other popular editors. Often with multiple suggestions, in case you don’t like the first solution it offers.

It’s no replacement for coding knowledge. You’ll still need to know how the plumbing works and it will get things wrong from time to time. But GitHub Copilot is almost guaranteed to make your coding jobs monumentally easier for only $10 per month.

When there are 50 applicants for every job, any marginal gain is worth having. Interviewsby.ai is a cracking little tool for perfecting your interview technique.

It asks you to paste in the job description of the role you’re applying for and then spits out a selection of questions you might face when being interviewed. The questions are well-targeted, although the more detailed the job description, the better the quality of the questions.

The key element is that it asks you to speak your responses as if you were in a live interview, which means you feel the same pressure of providing an instant response, as you would in the real thing. Your answers can last up to a minute, and when you’re done, the AI transcribes what you’ve said and provides feedback on those answers.

In my mock interview for a head of editorial role, I neglected to provide the specific examples from my previous experience that the question was asking for, which was a strong reminder to answer the question, not just spout what you think they want to hear. The feedback is generally encouraging and specific, although the suggested answers are less worthwhile, as the AI obviously doesn’t know your career history. Nevertheless, it’s a great, free training tool for anyone preparing to face an interview panel.

Video is critical on social media, but not everyone can afford a team to shoot content. Fliki could potentially save you a bundle on the camera crew.

Write a video script, choose a virtual host to narrate it, select stock video/images for each scene (or upload your own), and add some background music if you want. The AI voice narration sounds a little too synthetic, although you can manually alter the pitch of words or correct mispronunciations of brands or other words. But this will surely improve over time.

Fliki offers a free plan, so you can see how you get on with the tools, before subscribing to get rid of the watermarks and increase video quality. You’ll need a $21 per month standard account to hit Full HD quality, mind.

Notion has become one of those trendy apps; it’s a OneNote for people who like to be seen.

It’s one of those all-encompassing note-takers that can be used for creating to-do lists, bashing out ideas, taking notes from meetings, and even as a public wiki for your company’s products.

It’s recently added a set of very well-implemented AI tools to the mix. As you’re writing on any page, you can simply press the space bar at the start of a new line and activate the AI tools. Here, for example, you can ask the AI to autocomplete a passage you’re writing, list ten ideas on different ways you could market a new product, or create a to-do list of items you’ll need to pack for a trip.

If you’ve used it to take meeting notes, it can pick out the action points; if you’ve written a description of a new product line, the AI can take a stab at writing the press release.

The AI responses flow into the page within a couple of seconds and it’s easy to edit out any daft suggestions. “Adapter for electrical outlets” for my holiday to Crete? Don’t be daft, everyone knows you buy them afresh at the airport every time you fly abroad at vastly inflated prices.

Notion gives free account holders 20 chugs on the AI every month. If you find it invaluable, and you just might, then it’s $10 a month for unlimited usage.

Death by PowerPoint doesn’t only afflict an audience – it can drain a serious amount of time for the poor sod putting together a presentation too. SlidesAI wants to handle that for you.

This add-on for Google Slides invites you to copy and paste a large chunk of text that you want to make a presentation about. The free plan is limited to 2,500 characters, but the Pro plan ($10 per month) gives you 6,000 characters to play with and up to ten presentations per month. There’s also a $20 per month Premium plan.

I gave it an article I wrote about whether it’s acceptable to send an AI assistant to your virtual meetings on your behalf, and it did a solid job of picking out the key points, summarising them, and slotting them into an eight-slide presentation.

The AI also picks out the images, and here it wandered slightly off-piste. A slide on “AI’s meaningful contribution” to meetings was illustrated with a tulip, for instance, for reasons I simply can’t fathom. But a slide about Salesforce’s acquisition of Slack did come with a nice pic of Salesforce HQ. All the content can be tweaked, so if you’re in need of a quick-and-dirty presentation, SlidesAI could well be the AI you want to take to a meeting with you.

Microsoft hasn’t bet the house on AI, but it’s come close. The company has given Bing search and the Edge browser a new lease of life with AI, and now Microsoft has unveiled Windows Copilot. With Windows Copilot, AI is clearly being integrated heavily into Windows 11. So much so, that we’re a little surprised this isn’t the basis for Windows 12.

At its most basic level, Windows Copilot may stop you from ever having to dive into Settings or what’s left of the Control Panel again, with the ability to ask the AI assistant to turn on things such as focus modes or dark themes.

Windows Copilot also looks set to get some of the AI abilities we’ve listed in apps here, such as the option to drag across a PDF file and ask for a quick summary.

However, Windows Copilot gets really interesting with deep integration into apps, both from third parties and Microsoft’s own. A Microsoft demo video shows someone asking for help designing a business logo, for which they are directed to Adobe Express. Once the logo is designed, it’s exported back to the AI assistant, where the user asks it to be sent to members of a group in Microsoft Teams, which is performed automatically.

If Windows Copilot works as smoothly as it does in Microsoft’s demo, this could be a very big deal indeed. If it gets drowned in a sea of permissions prompts or favors Microsoft’s apps over all else, it could be the missed opportunity of 2023.

Source: https://www.itpro.com/technology/artificial-intelligence/amazing-ai-tools-to-try-today