Stepping onto the bandwagon before it fully departs, AI technology behind Google’s Bard chatbot is being embedded into Search, as announced at Google’s I/O conference, whose theme was all about how everything is being re-engineered to embrace generative AI. El Reg has neatly summarized the keynote here if you’re interested to learn more (“We sat through the Chocolate Factory’s PR blitz so you don’t have to”). Google announced copilot-like functionality for its cloud productivity suite while Microsoft unveiled the M365 Copilot preview that’s been running for a few months, is being extended further. Not all is rosy in the garden of AI, however. Distinguished scientist Geoffrey Hinton, a Turing Award winner and widely-described as “Godfather of AI”, has walked out of his Googly role amid concerns that AI will become sentient and enslave or kill us all. Interestingly, Hinton did not sign the Elon Musk-backed petition to halt AI development, effectively saying that if those currently working on it were to stop, others would pick up the baton. Microsoft’s chief scientist agrees.
The expanded M365 Copilot preview is a “paid-for” thing, and Microsoft’s Q3 earnings call did mention that Copilot will be additionally priced over and above whatever Office licenses a customer already has (though some AI related features will show up in E3/E5 licensed environments, such as the new Semantic Index which can be harnessed by Copilot but will be useful for giving more accurate search results even if Copilot is not in use).
Back in the present, there are some relatively new practical capabilities in both Bing AI and in the Edge browser’s discover feature, as discused in last week’s ToW. The Compose feature in the Sidebar lets you play with generating different types of written content, the kind of thing which will be integral to Copilot in all kinds of Office applications before long. The Insights tab on the same Sidebar gives you more info on the page you’re currently looking at, from a summary of the key points of the page, to some background on where that site is accessed from, how likely it is to be reliable and more. The core Bing AI search in a browser – in case you’re itchy about using Edge or its Bing Sidebar – has some new Another one of OpenAI’s groovy tools – Dall-E – generates images based on a text description, and Bing AI chat can feed directly into that. The image generation capability is now multi-lingual (with over 100 languages supported). It will also soon be possible to upload images to Chat, so you could ask it questions about what’s in the image. All free for now, but someday soon, we will need to pay the ferryman or the robot overlords will wreak their revenge. |
Tag: AI
680 – Edgy emails
The Edge browser has seen a lot of change in its life. Originally conceived as a successor to Internet Explorer, with its own modern web rendering engine and lots of additional features which are designed to complement the usage experience, like taking handwritten notes on top of a webpage or building a Reading List of pages or publications to come back to.
Later, the decision was taken to replace the browser with one based on a new core using Chromium, largely for reasons of compatibility and performance, but to carry on building new capability that would differentiate the new Edge browser from others that also use the Chromium rendering engine, including Google Chrome itself.
Recent builds of Edge have a Sidebar which includes a load of apps and integrations – the goal being that it can help multitask on the web by sharing complementary information or functionality alongside the page the user is looking at; think a shopping widget that would compare prices of the product on the page you’re viewing, showing where else you could buy that same thing.
A recent update to the Sidebar has been the inclusion of the new Bing search, which adds some very cool relevance capabilities that would allow you to fire the current page content straight at the Bing’s AI engine to summarize, rewrite or explain the contents – selected text, or the entire page you’re visiting.
Here’s an example of a reasonably detailed blog article (from early 2021) summarized into a few key paragraphs:
One of the more useful integrations in the Sidebar is the Outlook app (individual icons on the Sidebar can be enabled and disabled through the settings option; you can also dock other sites which will appear in the sidebar, though not necessarily with the context of the page you’re currently looking at). A recent – and somewhat controversial – change means that when you click a link from an email in desktop Outlook on a PC, it will open in Edge and the Outlook sidebar will be shown alongside, displaying the email that you clicked it from.
Once you’ve got the hang of this feature, it’s actually pretty cool – especially if the email is offering some context about what you’re supposed to be doing on that page, or if it’s a densely-packed missive full of clickbait and other nonsense:
Why is it controversial? Well, the point is that the extra functionality is happening due to the Sidebar in Edge, so clicking a link in Outlook if you’re using a different default browser wouldn’t have the same effect. Outlook, therefore, has decided to send links to Edge even if that’s not your usual browser, to the chagrin of some netizens. Be careful with doing things that annoy some people.
If you’d prefer that Outlook and Windows respected your choice to send all your links to a specific, non-Edge, browser, then it’s fairly easy (if not exactly easily discoverable) to set that. Go to File | Options | Advanced within Outlook, and look for the Link Handling option, and change it to Default Browser. This will mean opening the hyperlink in Chrome / Brave / Firefox / whatever, without the Sidebar doing its thing.
More change is on its way to Edge and Bing AI.
If you like Edge but would rather dispense with the Sidebar altogether, go to the “…” menu on the top right, select Settings | Sidebar and disable the Always show sidebar toggle.
You can use the same settings UI to play with other behaviours in the various apps that are pinned to the sidebar, too.
To add or hide apps on the sidebar, just show it, right-click on something and choose Customize sidebar, or use the “Add or remove apps…” feature from the Settings | Sidebar screen.
If you’d rather not to have the somewhat prominent Bing icon on the very top right of your Edge screen, look under the Discover section of this Settings UI, and if you flick the switch, the big blue b goes away.
674 – Here’s the (co)pilot
UK telly viewers in the early noughties may recall the surreal comedy show, Trigger Happy TV, with recurring characters like the aggressive squirrels or the guy with the massive phone (and that Nokia ring tone).
It was also known for some great soundtracks, like the fantastically titled Grandaddy song “He’s Simple, He’s Dumb, He’s the Pilot” (also used elsewhere). Tech news over recent weeks tells us that the pilot – or Copilot – is anything but dumb, even if it can be simple.
For Microsoft watchers, “Copilot” is a growing set of capabilities which are being built to add OpenAI functionality to other applications. With all the hoo-hah about ChatGPT and the generative AI that is now integrated into Bing (and available for everyone who wants it, not just early adopters), it’s easy to get different strands mixed up.
GPT-3 and now GPT-4 are the core language models which could underpin any number of applications’ use of what looks like artificial intelligence. ChatGPT is one web app built to hone some of the parameters of GPT-3 and put a chatbot front end to it. The new Bing and all the other stuff announced over the last few weeks is not using ChatGPT, but they do share some of the same technology underneath. Capisce?
There have been AI features aimed at making developers’ lives easier, such as Github Copilot (available since 2021), which uses another OpenAI tool called Codex, itself built to harness GPT-3. For developers on Power Platform, there have been AI functions for years too, though some capability has been recently added.
Everyday users of Dynamics 365 and Office applications will soon get Copilot capabilities to help automate boring tasks, like “work”. Do bear in mind that announcing something and making something available – in limited preview form or generally – are different activities. Copilot for Office apps like Outlook might be a few weeks or months away for most of us, but who can’t wait for AI to automatically read and reply to all their emails?
The future with our robot overlords never looked so appealing.
For a growing summary of Copilot announcements, see the hugely popular LinkedIn post from Jack Rowbotham.
668 – Ay Eye Ay Eye, Oh
Not Aye-Aye, Ally-Ally, nor Why-Eye, but Ay-Eye, as in A.I. And not the cheesy Spielberg flick. The tech news has been all about artificial intelligence recently, whether it’s ChatGPT writing essays or giving witty responses, to Microsoft committing another chunk of change to its developer, OpenAI.
Original backers of OpenAI include Tony Stark (who has since resigned from the board in order to discombobulate the world in other ways) and AWS, though Amazon has warned employees not to accidentally leak company secrets to ChatGPT and its CTO has been less than enthused.
ChatGPT is just one application – a conversational chatbot – using the underlying language technology that is GPT-3, developed by the OpenAI organization and first released over 2 years ago. It parses language and using previously analyzed data sets, gives plausible-sounding responses.
Further evolutions could be tuned for particular tasks, like generating code – as already available in PowerApps (using GPT-3 to help build formulae) or GitHub CodePilot (which uses other OpenAI technology that extends GPT-3). Maybe other variants could be used for interviews or auto-generating clickbait news articles and blog posts.
Another use for GPT-n has been unveiled in the New Bing – an old dog has maybe learned some new tricks?
You’ll need to join the waitlist initially but this could ultimately be a transformational search technology. Google responded quickly by announcing Bard, though Googling “Google Bard” will tell you how one simple mistake hit the share price. No technology leader lasts forever, unless things coalesce to there being only one.
Other AI models are available, such as OpenAI alternative, Cohere, and there are plenty of sites out there touting AI based services (even if they’re repainting an existing thing to have .ai at the end of it). For some mind-blowing inspiration including AI-generated, royalty-free music or stock photos, see this list.
613 – Ready for your close-up?
Outlook made some helpful suggestions in response – just one example of new AI functionality showing up by dint of subscription services.
Mirroring your video makes it easier to interact with your own environment while you’re looking at the screen, though it doesn’t affect what others in the meeting see. Lighting Correction is a one-setting tweak for fixing the contrast and brightness, which can be handy when it’s dark outside and your room lighting isn’t ideal.
Check out the Teams blog, and look forward to lots more new features arriving later in the year. Another tweak in managing your own video comes from users’ feedback, where they don’t want to see their own video window, thinking that it’s distracting when looking at a gallery of other attendees in a meeting.
Perfect for checking out how the facial buffing has worked out. |
502 – Presenting PowerPoint Subtitles
When they do it properly, you’ll often see presenters kick off by fishing
PowerPoint will typically be set up to use Presenter View by default, and the screen that’s being shared will be showing the speaker notes / next slides etc, while the full-screen content is being displayed on the 2nd monitor that isn’t being shared.
Choose the language you’d like to display, the location of the subtitles and when you start presenting, the machine will listen to every word you say and will either display what it thinks you’ve said in your own language, or it can use an online service to translate to subtitles in over 60 languages. It’s fantastic. See more here. Go and try it now.
Windows has a closed captioning setting page that applies to other apps that support it, too, if you’d like to show subtitles on video that has the content already defined. Closed Captioning is legislated by several countries, for traditionally-broadcast media as well as online video. You may also want to add captions to videos that you plan to embed – more, here. |
Tip o’ the Week 491 – PowerPoint layout tips
There are many resources out there to try to help you make better slides – from how-to videos to sites puffing a mix of obvious things and a few obscure and never-used tricks (eg here or here), and PowerPoint itself is adding technology to try to guide you within the app.
If you don’t see the Design Ideas pane on the right, look for the icon on the Design tab, under, er Designer. The PowerPoint Designer team has recently announced that one billion slides have been created or massaged using this technology, and they have previewed some other exciting stuff to come – read more here. A cool Presenter Coach function will soon let you practice your presentation to the machine – presumably there isn’t some poor soul listening in for real – and you’ll get feedback on pace, use of words and so on. Watch the preview. No need to imagine Presenter Coach is sitting in his or her undies either. When it comes to laying out simple objects on a slide, though, you might not need advanced AI to guide you, rather a gentle helping hand. As well as using the Align functionality that will ensure shapes, boxes, charts etc, are lined up with each other, spread evenly and so on, when you’re dragging or resizing items you might see dotted lines indicating how the object is placed in relation to other shapes or to the slide itself… In the diagram above, the blue box is now in the middle of the slide, and is as far from the orange box as the gap between the top of the orange box and the top of the grey one. There are lots of subtle clues like this when sizing and placing objects, and it’s even possible to set your own guides up if you’re customising a slide master. |
Tip o’ the Week 460 – AI, AI, Oh…
Of course, many of us know AI as a term used to refer to a host of related technologies, such as speech and natural language recognition, visual identification and machine learning. For a great example on practical and potentially revolutionary uses of AI, see Dr Chris Bishop’s talk at Future Decoded 2018 – watch day 1 highlights starting from 1:39, or jump to 1:50 for the example of the company using machine learning to make some world-changing medical advances.
One simple example is image searching – if you upload photos to consumer OneDrive (directly from your phone perhaps), the OneDrive service will now scan images for text that can be recognized… so if you took a photo of a receipt for expenses, OneDrive might be able to find it if you can remember what kind of food it was.
More AI goodness is to come to Office 365 and OneDrive users in the near future – automatically transcribing content from videos stored online (using the same technology from the Azure Video Indexer and Microsoft Stream), to real-time PowerPoint captions. Watch this space… and mind the robots. |
Tip o’ the Week 430 – developers, developers, developers
Apple has its WWDC in California every year (for a long time, in San Francisco), and at its peak was a quasi-religious experience for the faithful. Other similar keynotes sometimes caused deep soul searching and gnashing of teeth. The Microsoft one used to be the PDC, until the upcoming launch of Windows 8 meant it was time to try to win the hearts & minds of app developers, so //build/ became rooted in California in the hope that the groovy kids would build their apps on Windows and Windows Phone. Now that ship has largely sailed, it’s gone back up to the Pacific North West, with the focus more on other areas.
With Azure Sphere, there’s a way for developers to build secure and highly functional ways to process data on-board and communicate with devices, so they can concentrate more on what their apps do, and on the data, less on managing the “things” which generate it. For all of the breakouts at Build and the keynotes on-demand, see here. Back in the non-cloud city, Google has adopted a similar developer ra-ra method, with its Google I/O conference also taking place in and around San Francisco, also (like WWDC and Build) formerly at Moscone. It happened this past week, too. Like everyone else, some major announcements and some knock-em dead demos are reserved for the attendees to get buzzed on, generating plenty of external coverage and crafting an image around how innovative and forward thinking the company is. Google Duplex, shown this week to gasps from the crowd, looks like a great way of avoiding dealing with ordinary people any more, a point picked up by one writer who called it “selfish”. Does a reliance on barking orders at robot assistants and the increasing sophistication of AI in bots and so on, mean the beginning of the end for politeness and to the service industry? A topic for further consideration, surely. |