HT Wired Wisdom: ChatGPT, Alexa and Copilot would like to have a conversation

These AI chatbots are learning how to talk! You’ll, at some point in the next two weeks, be able to speak with OpenAI’s ChatGPT chatbot. It’ll reply, the same way. Five different voices to choose from. At least from my limited sample interaction, Sky’s voice seems most soothing. The chatbot will also be deciphering images with context you provide – upload an image and ask ChatGPT what it is you need to ask (analyse a graph, identify a flower or figure out complex controls on an appliance). Mind you, voice chats will be limited to ChatGPT’s apps for Android and iPhone. Image search shall be available across platforms, including desktop web browsers. Paying subscribers on a Plus (that’s $20 a month; or around Rs 1,663) or Enterprise plan (customised, for businesses), will get first access.

At work here is a neural network (specifically an open-source speech recognition system) called Whisper, which OpenAI has been working on for a while now. There are two parts to this. The first task is – transcribe the words we speak to the chatbot, into text, for it to understand. Once that’s done, the new text-to-speech model is capable of generating human-like audio from text. Or even a few seconds of sample speech that it hears.

It has all come to fruition, and it should be no surprise to anyone how human-like Sky, Juniper, Cove, Ember and Breeze sound. Just last week, I had pointed out how Amazon’s Alexa and similar voice assistants had transitioned from being considered a futuristic, to now a basic tool everyone expects at their beck and call on a phone, a smart speaker or even a smart TV.

New Echo and Fire TVs aside, generative AI is the big leap for Amazon’s Alexa

It is no surprise, Amazon is looking forward to enabling the biggest update for Alexa voice assistant, in years. That also revolves around voice, conversations, context and accuracy. A significant step forward to a smart assistant experience that has been refined over the years, but remained static in many ways. Capability and scope included. Quite when Alexa gains the new conversational capabilities globally, remains to be seen. The new updates rolls out later this year, in the US only, for now.

But what would you make of a tool that can create an extremely realistic sounding virtual voice, with text inputs or a few seconds of listening? This could open doors to potentially a lot of misuse. We’ve all already seen the dangers of fictitious content on social media, generated by equally powerful text-to-image tools. OpenAI is very clear access to this tool will be controlled to specific use cases (such as this) and limited partners (Spotify’s using this for their voice translation pilot).

Not the only AI chatbot that likes to talk. Apparently, Meta is working on one too (you might hear about it by the time you read this on Thursday morning; in which case, we’ll have a conversation about it next week). The core audience will be the younger aged demographic. Apparently, one that’ll have a “sassy” side to its personality.

You may like to read…

(Premium): Look who’s talking: Vishal Mathur on how AI chatbots learn

(Premium): Spot the differences: How is AI art getting so much better, so fast?

BENCHMARK

If any of you were expecting anything on the lines of a revolutionary new iPhone, you were walking down the wrong path from the outset. Long gone are the days of wholesale changes, year on year, with new smartphones. We’re already so far ahead with the pace of evolution (even mid-range smartphones are more powerful than most PCs), there is no need to start with a blank canvas every year. Incremental steps forward, if done with some thought, can be the collective forward step that can define an upgrade. The definition of which is subjective. A new phone may be better in every way for you, worthy of splurging money on. That’s not how it may be, for me.

That conundrum leads us to the Apple iPhone 15 Pro Max. This was never going to be a big refresh year (that’s likely to be 2024, or so we hope), but even with that as the foundation, I’ve called this a pivotal moment. Similar to the iPhone X, which in 2017, marked a new approach. This year, there are shades of the same. Titanium frame, for example, isn’t just a material switch to make the spec sheet look fancy. That, alongside a recycled aluminum skeleton (and at 221 grams) make this the lightest iPhone Pro Max, in a really long time. Also, lighter than the Samsung Galaxy S23 Ultra (234 grams).

Apple iPhone 15 Pro Max review: Many small steps together compose a giant leap

Often, the little things do matter. Unfortunately, not many pay much attention to. The Action Button, for instance. One could always argue no one really wanted that changed. But it isn’t just a design change, from a toggle to long press. It can be configured for more actions (quickly open an app, launch the camera or enable Translate, for instance), which may simplify certain tasks for many users.

There are external forces that have shaped this iPhone’s move towards a global standard that is fast evolving. Thank the European regulators for that. You’ll realise the convenience of charging cable simplification, soon enough. But trust Apple to have two distinct implementations of USB-C’s capabilities, for the sake of differentiation between its consumer line (iPhone 15 and iPhone 15 Plus) and the Pro phones. We’ve detailed the speed differences (and the big miss with the bundled cable), in the review.

There are two sides to the coin signifying photography improvements. Megapixel counts remain the same, but Apple insists there are generational sensor improvements. The step up to 5x optical zoom, from 3x, can always be argued is still lesser than some flagship Android phones. Yet, wouldn’t you classify this as a successful upgrade if a 5x optical zoom photo on an iPhone 15 Pro Max retains more useful detailing than Google Pixel, and matches the Samsung Galaxy S23 Ultra has more megapixels to work with? Remember I’d mentioned finer details – that’s the photography pipeline which Apple has optimised, rather than simply add more megapixels on the three cameras. It’d have been easier, and nicer to look at on the spec sheet.

But if you still feel that the iPhone 15 Pro Max’s camera troika is inferior to the Galaxy S23 Ultra or the Pixel 7 Pro, think of this as Michael Schumacher constantly wringing performance out of a decidedly inferior Ferrari in the late 1990s against a faster McLaren Mercedes driven by his on-form rival, Mika Hakkinen. Numerical advantage doesn’t matter, often.

PLATFORM

Are we finally witnessing an emergence of rivals, to challenge the monopoly of the often criticised Google Play Store? Digital payments platform PhonePe has confirmed that the Indus Appstore for Android phones will be launching soon, with support for 12 Indian languages. First things first, the Indus Appstore Developer Platform is now live, which means Android app developers can register and upload their apps in preparation for the Indus Appstore’s launch.

PhonePe says app listings will be free for the first year, and a “nominal annual fee will apply” from the second year onwards. That said, no commission will be cornered for in-app transactions or payments, and developers will have the option to use any payment gateway or methods, within their apps. It’ll be interesting to see how this made-in-India application store evolves in the coming months. And how many phone makers go about pre-loading this as the application store (or even as a choice alongside the Play Store) on Android phones they sell.

Also read…

Google looks to Android’s future, with a wary eye on the past

Google pauses Play Billing changes in India, for now

[Exclusive] As it battles Google Maps, MapmyIndia banks on patience and expertise

INTELLIGENCE

Just a sign of the times we live in. You’ll just have to bear a bit longer with our AI conversations. By the time you read this, a new Windows 11 update would be available for your desktop, laptop or convertible computing device. I am not certainly not wasting your time talking about a run of the mill monthly update with some fixes and an occasional feature addition. This is, instead, the big one. Windows Copilot, an AI assistant which was first demoed over the summer, will finally be arriving to your computer.

Microsoft’s next Windows 11 update adds AI at every step on your path

Copilot in Word. Copilot in Excel. Copilot in PowerPoint. Copilot in Outlook. Copilot in OneNote. Just an illustration of how deep this layer of AI infusion will be. Not to forget, the chatbot that’ll reside on the taskbar, capable of also managing certain Windows 11 options and settings. It’s already at home in the Edge browser; that’ll continue too. Still in the preview stage, but now that it’ll be available on millions of Windows 11 PCs, uptake will likely be tremendous. So is the potential for utility. I wouldn’t mind a summary of a lengthy Word document or an email thread (press releases sometimes tend to be that way). That’s just one example of what Copilot can do.

Windows 11 updates have a wide-scale impact since it is reaching millions of users globally. You’ll be relieved to know a redesigned File Explorer is arriving with this update too. Much needed, we’ve waited a long time. Outlook’s update gets streamlining of calendar and mails across multiple accounts. Paint, Photos, Clipchamp and Notepad get significant improvements too. As does the Bing Image creator – this AI text to image creation tool will now use use OpenAI’s DALL.E 3 model, which should give it better context and more knowledge to work with.

Would you now agree, this is one Windows update, certainly worthy of a lengthy discussion?

Were you forwarded this email? Did you stumble upon it online? Sign up here.

Written and edited by Vishal Shanker Mathur. Produced by Md Shad Hasnain.