Before we get to our stories, coincidentally leading with one on digital humans used in customer service, we want to confirm that we did indeed hear from Air Canada, who confirmed that the chat-bot that got them into the recent lawsuit was NOT generative AI. I received a tip from someone who should be a very reliable source that the chat-bot was actually “shadow IT” – not implemented by the corporations IT department where it would have presumably received a controlled testing and implementation. Once again, I’ve given Air Canada the right to comment on this information we received.
Whether we like the idea of AI driven service agents or not, we should be making our decisions based on the best evidence available.
And now, back to our regularly scheduled programming.
Deutche Telekom is using what they call “digital humans” for customer service and claiming great results. Mark Zuckerberg gives some insight into Meta’s different approach to AI strategy in recent interviews and Elon Musk suffers quality issues on two offerings which may have been rushed to market.”
All this and more on the “Murphy was an optimist” edition of Hashtag Trending. I’m your host, Jim Love. Let’s get into it.
First up, the next time you call customer service for your mobile provider, you may end up speaking to a digital human powered by artificial intelligence instead of a real person.
Deutsche Telekom, the parent company of T-Mobile in the U.S., has started using a digital human representative named “Max” for some customer interactions. Max is a computer-generated character that looks and sounds lifelike.
Simon Grieve, an executive at the company that created Max, explains why they opted for this AI approach: “Max looks very real. He’s obviously CGI because we don’t want to trick people. But he’s very high quality. We really focus on being as human-like as possible but without feeling creeped out.”
Unlike a traditional chatbot, Max can make natural gestures and facial expressions to try to build rapport with customers. Deutsche Telekom says Max has been successful so far, resolving 10% more issues than their chatbot alone.
However, you can still request to speak to a human agent if the digital human can’t resolve your problem. Telecom analysts say this blend of AI and human service could become more common as companies look to reduce costs while maintaining person-to-person interaction.
These AI service agents are getting better and better. Are they ready for “prime time” yet? Major companies like Canada’s Canadian Tire are implementing their own customer service applications, in what appears to be a very controlled manner.
One of our sponsors has a sample of bot using generative AI in text and soon voice at padv.ca. And the bot referred to in this story can be viewed at fittingly, digitalhumans.com
Sources include: Fierce Network
Elon Musk’s new AI tool called Grok is making headlines again – but not for the reasons he hoped. The tool, touted by Musk as the “anti-woke AI”, has been generating bizarre fake news stories based on jokes posted on Twitter or X.
One of the key issues may be that Grok isn’t only “anti-woke” – it doesn’t really cope with humour and satire.
In a story about the recent eclipse, Grok published a news item to users with the alarming headline: “Sun’s Odd Behavior: Experts Baffled.” It claimed the sun was “behaving unusually” with “no official explanation from scientific authorities.”
The confusion stemmed from Twitter users making jokes about the sun’s strange appearance, like this one from comedian Michael Ian Black (clip): “WHAT THE —- IS HAPPENING TO THE SUN RIGHT NOW?????”
Grok seemingly took the jokes as face value and invented a baffling news story, just a week after falsely claiming Iran had launched missile strikes on Israel based on unrelated tweets.
While Musk has praised Grok as superior to “woke AI” from Google and others, this latest flub shows generative AI still has a long way to go before outputting credible, factual information.
So Twitter’s paid subscribers got an AI-generated fake news story this week instead of…well, the actual news.
Sources include: Gizmodo
In the intensifying race to develop powerful artificial intelligence, Meta CEO Mark Zuckerberg is, perhaps by necessity, taking a different approach.
While AI frontrunners like Google, OpenAI and Microsoft are locked in an arms race to amass more training data, Mark Zuckerberg is taking a different direction. Having reportedly flirted with the idea of buying a publishing company or dealing with the lawsuits and copyright claims that are hitting OpenAI, Zuckerberberg is talking about another concept – he calls it “feed-back loops.” Instead of focusing on absorbing vast amounts of data, he’s focusing on how people use and interact with data – something that Meta is perfectly poised to do.
Zuckerberg says “The thing that I think is going to be more valuable is the feedback loops rather than any kind of upfront corpus…Having a lot of people use it and then seeing how people use it and being able to improve from there is actually going to be a more differentiating thing over time.”
But Zuckerberg’s focus on iterative feedback over initial data is about more than just refining Meta’s AI capabilities. He’s outspoken that he doesn’t want AI development to end up controlled by a few “gatekeeper” companies like Apple and Google have over mobile apps.
Zuckerberg criticized how those companies can dictate what apps and features get released. He said he doesn’t want any other tech giants telling Meta “what we’re allowed to build” when it comes to AI.
That’s part of why Meta open-sourced its powerful language model Llama – to avoid having to get approval from closed systems owned by competitors. Zuckerberg wants feedback loops from widespread public usage to rapidly improve Llama, rather than being boxed in.
So while rivals focus on simply ingesting more data upfront, Zuckerberg is betting on what might be a more grassroots feedback-driven approach he believes will prove more valuable long-term.
There may be some pitfalls along the way. There are already reports that you can’t shut down Meta’s AI, at least not easily, and people may react to the learning that is going on from their interactions and behaviours, but Meta has confronted these types of issue in the past and so far, they are have survived the challenges as people seem willing to give up their data and even privacy for free access.
And this strategy may indeed give Meta more control over its AI destiny amid rising competition from Microsoft, Google and others.
Sources include: Business Insider and Analytics India
And I promise that I’ll try not to do more stories on the Tesla Cyber Truck, but with sales softening and Musk reportedly now talking about a 20 percent reduction in Tesla headcount, he may want to face the reality that quality may be one of his issues.
The Cybertruck was billed as an indestructible machine – “literally bulletproof” in Elon Musk’s own words. But it seems even something as mundane as a car wash can turn this six-figure truck into an expensive paperweight.
One Cybertruck owner on TikTok says after taking his truck through an automatic wash, the massive center screen that controls critical driving functions like speed and the rear camera went totally black.
Following Tesla’s recommendations for a hard reboot didn’t immediately solve the issue. Only after waiting five hours did the screen finally reset, with the owner joking “where can I get a computer that resets in five hours for work?”
It highlights the potential downsides of vehicles grown increasingly reliant on computer systems and software. What was once as simple as a garden hose now requires putting the Cybertruck into a special “car wash mode” according to the owner’s manual.
The manual also cautions “do not wash in direct sunlight” and says “failure to use car wash mode may result in damage” that wouldn’t be covered by the warranty.
It’s not the only pitfall badgering the Cybertruck’s rugged image. Owners have reported rust forming on the stainless steel body, which Tesla says stems from failing to quickly clean off corrosive substances.
There have also been issues with the truck’s accelerator pedal, prompting a recent safety recall from regulators just months after customer deliveries started.
So it’s just a thought that maybe instead of staff reductions, Tesla might divert some more resources into design and quality control? Of course, what do I know, I’m not a genius or a billionaire.
Hashtag trending goes to air five days a week with a weekend interview show. And we are also on YouTube.
Find us at our new home at technewsday.ca or .com – you pick. And you can reach me with comments, suggestions or even criticism at therealjimlove@gmail.com or at editorial@technewsday.ca
I’m your host Jim Love, have a Terrific Tuesday.
