Babelfish is Here: CES 2025 Kicks Off with Real-Time Translation Earbuds. Need high speed internet? Fly United. And has Artificial General Intelligence already occurred.
Welcome to Hashtag Trending. I’m your host, Jim Love. Let’s get into it.
This Tuesday we have another special edition of our interview program Afterword. Today we have Zainul Mawji, President, TELUS Consumer Solutions and she’ll be talking about an issue with Canadian government regulation. If you’re interested, just stay on after the closing credits.
Babelfish is Here: CES 2025 Kicks Off with Timekettle’s Real-Time Translation Earbuds
CES 2025 in Las Vegas opened with a nod to science fiction made real, as a company called Timekettle unveiled the W4 Pro earbuds—a modern take on the Babelfish from Douglas Adams’ The Hitchhiker’s Guide to the Galaxy and Star Trek’s iconic Universal Translator. These groundbreaking earbuds aim to break language barriers by providing real-time, two-way translation for phone and video calls.
The W4 Pro earbuds translate phone calls automatically in 40 languages and 93 different accents with a latency of just three to five seconds. This near-instant cross-lingual communication works seamlessly with any telecommunication app or traditional phone system. Users can even share one earbud with someone else to facilitate face-to-face translations, maintaining the same low-latency response.
Powered by Timekettle’s Babel OS, the earbuds enhance the translation experience by anticipating dialogue, offering customizable lexicons, and capturing human emotion and tonality—making the dream of universal communication closer to reality. The Babel OS update will also be available for older Timekettle models.
Beyond translation, the W4 Pro functions as standard Bluetooth earbuds for music and calls. They offer six hours of continuous translation and an extra hour of use after a quick 10-minute charge. Priced at $449, the earbuds are available now on Timekettle’s website.
United Airlines Accelerates Starlink Rollout for Inflight Connectivity
It has seemed that every damn perk you get in a flight is an additional cost. From luggage to seat selection and even to use the stupid (and often painfully slow) internet.So it was a bit of a surprise when United Airlines announced that it would fast-tracked the rollout of SpaceX’s Starlink internet service across its fleet, aiming to bring high-speed connectivity to passengers as early as spring 2025. United plans to begin testing Starlink next month, with its first commercial flight using the service on a regional aircraft expected by spring. The airline will complete Starlink installations across its entire two-cabin regional fleet by the end of 2025, with its first mainline aircraft going online by year’s end.
The service will be free for MileagePlus members and offers uninterrupted streaming, gaming, shopping, and inflight entertainment from gate to gate. And MileagePlus is free to join, with members can use miles for almost any purchase, including seats and onboard services.
United signed a deal with SpaceX in September, marking the largest Starlink agreement in the aviation sector. The airline aims to become the first U.S. carrier to offer high-speed, low-latency internet across its entire fleet, ensuring a seamless inflight experience on both seatback screens and personal devices.
United CEO of MileagePlus Richard Nunn said “Adding Starlink is going to revolutionize the experience of flying United,”
Are we already at Artificial General Intelligence?
Yesterday, Sam Altman, CEO of OpenAI teased with world with a tweet:
i always wanted to write a six-word story. here it is:
___
near the singularity; unclear which side.
It’s clear that the company is moving fast towards a future that may go beyond AGI, the idea that AI can be as smart or even smarter than a well educated person. It continues to setting its sights on what it calls “superintelligence” — AI systems that massively surpass human intelligence.
The description of that event has borrowed a term from physics – the singularity – which essentially means an event or impact that cannot be measured.
It’s the stuff of science fiction novels, but most experts believe that it’s in our immediate future.
But with that cryptic phrase, Altman has acknowledged that Artificial General Intelligence, where AI is smarter than a highly educated person may have already been achieved. He said as much is a subsequent blog post:
“We are now confident we know how to build AGI as we have traditionally understood it. We believe that, in 2025, we may see the first AI agents “join the workforce” and materially change the output of companies. We continue to believe that iteratively putting great tools in the hands of people leads to great, broadly-distributed outcomes.”
But have you ever wondered why Altman is always being so cryptic? Is o3 AGI or not? Don’t be cute about it.
But there may be a really good reason.
OpenAI is in a pitched battle with Elon Musk, Mark Zuckerberg and others to be the leader in Artificial Intelligence. Musk, a former friend and ally of Altman, is now doing everything in his power to hinder OpenAI’s progress. He’s tried to lockdown Nvidia’s production of its new GPU chips – and Musk has the money to do that.
OpenAI, starting as a non-profit and trying to find a new structure where they can raise the tens and may hundreds of billions of dollars to keep moving forward. Despite its relative commercial success, with a growing number of corporate clients and retail subscribers, OpenAI is rumoured to be losing close to 45 billion dollars in this year alone. And there is more to invest.
But OpenAI has an “ace in the hole.” Microsoft, its partner and a major funder is investing 80 billion dollars this year into data centres and production facilities.
BUT – Microsoft’s deal with OpenAI has a limit. When Artificial General Intelligence is achieved, Microsoft loses its rights to access OpenAI technology.
In the original agreement, the definition of AGI was left “open” to OpenAI to determine. Now, according to a leaked document, Microsoft and OpenAI have agreed that AGI would occur when it accounted for 100 billion dollars a year in profit. So that’s why Altman may be playing cutesy with the definition. But that doesn’t mean that AI won’t have an enormous impact long before that occurs.
Altman is suggesting is that these AI agents will begin to materially change business outputs as early as this year.
OpenAI’s latest model, known as o3, is raising eyebrows. The o3 model achieved breakthrough results on the ARC-AGI test, which measures how well an AI can adapt to entirely new problems without relying on past training. Unlike traditional tests that focus on recalling facts or following programmed instructions, ARC challenges the AI to figure out the rules behind puzzles it has never seen before. Imagine being shown several examples of shapes moving or changing color based on hidden rules, then being asked to predict the outcome of a new puzzle. For humans, this type of reasoning feels natural. But for AI, it has been a major hurdle — until now.
The o3 model’s success on ARC suggests it can generalize knowledge and solve problems in ways previously thought to be unique to human intelligence. This is a critical step toward AGI. However, there’s still a significant gap to close when it comes to efficiency. The human brain operates on just 20 watts of power, roughly the energy needed to power a light bulb. In contrast, running a model like o3 requires thousands of watts, and on a single task may use enough power to run an entire household for two months.
But if you express that in terms of cost, you might see the problem differently.
Or if you express it in dollar terms, a frontier model costs between $17 to $20 per task in its low-compute mode.
While that may seem expensive, it highlights how AI is becoming competitive in tasks that require advanced reasoning and problem-solving, though with a much higher energy footprint. How many complex problems do humans solve per hour?
O3 may solve another problem. We are running out of data to train these new models. Not only because the vast amount of data that is available has already been used, but there are also legal issues and new barriers making getting what little new data they can find more difficult and more expensive.
But a new model, that can learn from situations it has not been trained on, would constitute an incredible breakthrough. And in last month’s 12 days of shipmas, Altman and crew showed how they could already fine tune a model with as little as a dozen new prompts.
Here’s the bottom line. Altman may be afraid or limited in what he can call this new development. But the impact will be large. On all of us.
And that impact is coming faster than we might think.
Near the singularity, it’s unclear which side we’re on.
That’s our show for today. Remember, stay tuned for Afterword, our interview show. This week we have Zainul Mawji, President, TELUS Consumer Solutions,
You can find links in our show notes at technewsday.com or .ca, take your pick. You can reach me with comments, questions, or tips at editorial@technewsday.ca. I’m your host, Jim Love. Have a Terrific Tuesday.
