HT Wired Wisdom: Not all (humans or tech cos) have the courage to do the right thing
For some, it comes as quite easy to be hypothetical. Hallucinatory. Have fancy ideas about an alternate reality. At the same time, being conscientious, ethical and principled, aren’t easy paths to choose. It is true for people, and it is true for tech companies. There are often different reasons why I may say something on these lines. This week, it is in recognition of the tough, righteous path Google has consistently held on to, with regards to artificial intelligence. This isn’t to say others have dropped the ball (though some have; hence the flurry of generated images passing off as real ones on social media), but Google made the sensible choice despite making little business sense. Or for perceptions. And this week as testament to the good results from being conscientious.
Responsible AI usually isn’t up debate at Google. Usually. That is, until a somewhat dodgy demo video that attempts to show off Gemini’s multimodal capabilities came about. If I point you to this video (here on YouTube), do read the description text. “For the purposes of this demo, latency has been reduced and Gemini outputs have been shortened for brevity.” All of a sudden, promises of Gemini’s multimodal capabilities on which these seemingly remarkable conversational skills would built (or as the video tries to illustrate), isn’t at all that anymore. You still cannot have a conversation with a chatbot, as you would with a human. Not to say Gemini may or may not develop to that extent, but heavy editing makes it seem more capable than it presently actually is. Suddenly, it isn’t as superior to ChatGPT, as the video first tries to show. I have faith this is just a blip, in an otherwise sensible approach.
Quite a contrast nevertheless, from generally restrained and careful steps that Google has thus far taken with AI. Something I’ve often appreciated in our previous conversations through the year. Google took their own time with the Bard generative AI chatbot, when it finally released a few months ago to general public. After a test version was teased over the summer. Despite intense pressure, considering this public access unlocked almost a year after OpenAI’s ChatGPT, which was incredibly popular by then. Many mocked Google Bard as an afterthought, not that it mattered to the tech giant. Being responsible and sensible, paid off. Thus far. Despite this too-clever bit of video editing, Bard may still be primed to become the smartest chatbot out there, underlined by what’s still potentially the smartest AI model at this time.
Facing fire
While their competition has just surpassed them, OpenAI’s still resetting their house in some sort of order. An upheaval, caused apparently, for no good reason. Helen Toner, one of the board members who fired Sam Altman and then resigned when Altman returned, is still unable to give a specific reason for why the CEO was fired in the first place. Except, “our goal in firing Sam was to strengthen OpenAI and make it more able to achieve its mission,” which is what she said in an interview with WSJ. That tells us absolutely nothing new. Or perhaps, it indeed says a lot about a board that presided over everything going pear shaped.
DISPATCH
Apple doesn’t want Android users to access iMessage. That’s been very clear, for years. We can criticise it all we want, but the fact is quite simple – it is Apple’s choice, what they want or don’t want to do with a service they’ve developed. Privacy is their concern, and thus, that’s the reasoning they stand on when arguing it’s exclusivity to Apple’s own devices. Fair enough. Because Android phone makers and third-party app developers have exhibited extreme levels of immaturity over the past few weeks in the sudden re-emergence of an ill-advised desperation to get iMessage running on Android. Somehow. Data privacy be damned.
Our regular readers would be clued in with our conversation about how phone maker Nothing allowed itself to be taken for a ride by app developer Sunbird’s promises of bringing iPhones and Android phones closer together. Nothing launched Nothing Chats and even said sorry to Tim Cook for having achieved what many thought wasn’t possible. It was a particularly bad week for tech, and this added to the kerfuffle. None of this stopped another app developer, Beeper, from attempting to set the world right. And it has been an absolute mess. First there was the joy of having found, in their own words, a solution that’s essentially “because the iMessage protocol and encryption have been reverse engineered by jjtech, a security researcher.” That alongside insistence that no ‘Mac Mini farms’ are being used in this process and all messages are end-to-end encrypted (quite how, no can ever be too sure). Another one of the red flags, claims that no Apple ID is needed to get these to work. I wouldn’t trust my communication though a route that’s been reverse engineered by a dude with a laptop. It is bound to get shut down.
Clearly, neither does Apple. And Apple is shutting this down. Rightly so. To give you a simplified run-through of how this works – this dude’s big breakthrough idea is to somehow tap into Apple’s iMessage protocol that’s used to deliver all messages and convince Apple’s servers that’s indeed an Apple device from where the message is originating – which it isn’t (it’s an Android phone) and thereby use the “fake credentials” that Apple’s referenced in this time, as a concern they aren’t willing to tolerate.
I’d often praised Beeper for their original app that attempted to put a bunch of simpler to tame messaging apps within one interface, which is great news for the messaging and notification overload we often expect. It is called Beeper Cloud, by the way. But this, the Beeper Mini, is just not it.
Beeper may try its best to get Beeper Mini (that’s $1.99 per month, mind you) up and running again, but Apple will keep cutting off these nefarious methods. “These techniques posed significant risks to user security and privacy, including the potential for metadata exposure and enabling unwanted messages, spam, and phishing attacks. We will continue to make updates in the future to protect our users,” an Apple official statement reads. One of those will be iMessage Contact Key verification with iOS 17.2 update.
Shifty tactics and a vicious attitude, never work.
领英推荐
ECOSYSTEM
Sometime ago, we had chatted about the privacy focused (aren’t these rare now, in the truest sense?) software company Proton upgrading its cloud storage service with apps that made it a worthwhile alternative to Google Drive or Microsoft OneDrive or even Dropbox, if you’re looking for a more robust, encrypted storage option. Alongside updates for Proton Pass with a high-security Proton Sentinel programme and updates for Proton Mail on Android and iPhone too. Proton’s ecosystem is becoming even more viable, with the VPN and Mail, two cherries (yes, the bonus of privacy) on a proverbial cake. This week, another update is rolling out. Specifically, for the Android app (phones and tablets, of course), that gives it more to work with as a photo and video backup solution.
But then again, Proton told me they’re looking to develop this into a replacement for an Android device’s gallery or photos app. For that, there are detailed controls to view and manage photos you’ve uploaded. But in the short and long term, if that’s really the vision, it’ll need to also integrate more features – photo editing, to be fair, is crucial. Some tools, just to get things going. And that’ll prove its biggest challenge, convincing people beyond the encryption and data security aspects. Because for Android phone users, Google Photos more often than not is the backup and editing destination. Let us also not forget the sense of inertia that comes with cloud storage choices and switching.
Hold that thought
There is a video I’d like you to watch, something this creator has generated using a gamut of AI tools (It’s here). From the looks of it, this clearly is a lazy Sunday in New York City. Look carefully, someone crossing without a care in the world has a close shave with the garbage truck (is AI smart?). The creator says generative models including Midjourney, PS GenFill, After Effects, and Magnific AI have been used to put this together. Perhaps a small step forward for generative AI and video? Nevertheless, the lo-fi music soundtrack is worth leaving this on, while you quickly refocus on work.
GEMINI
Touching back for a moment on the Google Gemini conversation. There are three models, that make up Google’s Gemini portfolio. In different sizes, for varying applications. And costs that inevitably go with them. As I pointed out, Google isn’t wasting much time in deploying any of the three models in some its own products, and also make them available to developers as well as enterprises. No surprise that Bard, in the face of competition from ChatGPT, Microsoft’s Copilot and Anthropic’s Claude, gets the medium sized Gemini Pro as its new foundation.
Our regular readers would remember the “What I asked the chatbots” segment from last week. It only made sense to ask Gemini (to be specific, the mid-sized Gemini Pro model) the same question. The response is longer, more detailed, even a bit philosophical at times. It illustrates those differences, intended improvements. Better context, detailed responses and the tone of conversation is more likeable too. Google didn’t spend millions in research to make Bard only more pleasant to talk to, but these differences indicate a smarter model that’s working out what you’ve asked.
The other immediate implementation is with the Google Pixel phones, the very impressive Pixel 8 Pro to be specific, at this time. An Android update is already rolling out (this is the Google Pixel December update; extra goodies for the Pixel 8 Pro phones), which brings the smallest sized Gemini Nano model – you should have this available on your phone, by the time you read this. Summarise in Recorder, Smart Reply in Gboard, just some initial functionalities. More phones, more apps, likely to be covered soon. It’ll really shine though when you use Photos for Photo Unblur or make document scans looks better than they otherwise may have.
Were you forwarded this email? Did you stumble upon it online? Sign up here.
Pensionato presso Nessuna
1 年Good afternoon