Beep boop - this is a robot. A new show has been posted to TWiT…
What are your thoughts about today’s show? We’d love to hear from you!
With AI in Microsoft 365, that is a real problem for us. Our users all have Office installed on their machines and use them every day. For regulatory reasons there is currently an AI ban in the company, no use of AI, with 1 exception, DeepL.
We have to come up with an AI Ethics document and AI Policy, which need to be approved by independent audiots, before we can use any AI - and after the policy, the AI steering group has to test and provide a business use case, before we can enable access to an AI - currently, most of the big AIs are blocked off by Firewall policy - but M365 Copilot has no way to shut it off, but using it at the current time is a disciplinary offence…
I use some AI privately - I used Perplexity yesterday to answer a couple of questions and the answers were verifiably correct, which makes a pleasant change. But I still haven’t found a real use for much AI. I played with Midjourney, but the results were so bad and I had no actual need for the finished images anyway, that I deinstalled the app again from my phone.
1 Like
Slight preface: It might all be Leo playing devil’s advocate. If he is, he deserves some kind of recognition.
Hmm. With this episode (and the new US administration and their new tech agenda), I fear that Leo at times has gone pretty far into the TESCREAL tech-bro deep end. I can even understand it: when you’ve dedicated your professional life to tech and then the tech oligarchy takes over the country - how could you not?
But, I have to say, the position that AI is now the evolutionary step of humankind (around 2/3 of the podcast) and that it’s awesome to now finally push forward with the new, rather extremist, and agitated political overlords in more or less a single-party system (all three branches in alignment) in charge for at least four years simply seems like a bizarre fever dream from the outside. Especially because the biggest two reasons seem to be “If we don’t do it, China is going to do it!” and “It is going to be interesting!”
Each scenario taken by itself -
- single-party-rule for at least four years of any kind of party (and not unlikely subsequent)
OR
- an unleashed tech oligarchy that can only demonstrate that it fills its own pockets and not really show that it generates benefits for society except for allowing for “free speech” (which really is a terribly cheaply placating good, when you think about it) is, at least to me, a source of great concern for the whole world. The US has both now. Plus an administration that seems hell-bent on pursuing weak minorities, destroying institutions, curbing the rule of any resemblance of law (lest simply curbing the law itself), filling one’s own pockets, flirting with national socialism, the list goes on…
Listening to TWiT, this week, seems like listening in to a group of people who try to come to terms with reality - each in their own way. Not that I’d or anyone on planet earth would have any idea of where to go next, but I found it really difficult to bring together the spooky impression that the past week of Trump-Tech-Bro rule left with me with Leo’s seemingly (at least to my ears) bubbly and optimistic take. As long as TikTok does well…
Ah, well. Let’s evolve now and have those die out that could not adapt. Can’t go quick enough. Otherwise, we’re like those old loom-hating good-for-nothings. While we’re at it, let’s serve those to the dogs that might need our help. It’s evolution time!
I understand that two hours of hand-wringing don’t make for good entertainment. I understand that it can be a challenge to report on and discuss this. But having a 120 minutes of fun and tech optimism in my ears just creates the strongest cognitive dissonance in a while - especially after such a week.
I don’t know which other old white man (let’s face it - and I’ll be one in a few decades, too) Leo went on a walk at the beach with, but the outcome seems to be a less reflective and differentiated (role played by?) Leo. I really hope, the other Leo comes back.
Might be up to advertising money - and I really hate to even think that. But I could understand. Times are tough, evolution is coming, and I’d also rather be on the good side of it… :S
PS: I partly write this since I heard Leo say that he likes his panel to disagree with him to come up with a good discussion. Maybe it’s all a role to rile up push-back from the panels - but man, people don’t push back too hard nowadays. Thanks to Sam for pushing slightly and carefully as far as I remembered. But there’s no Stacy (as a regular) on the network, any more. Thank god for those panellists that actually push back - THANK YOU BENITO, come to think of it!
2 Likes
I have several friends and acquaintances in the USA who are LGBTQ+, after last week, I am very worried for them. One I contacted said she was scared and looking to get to a “safe place”. I wish her and everyone like her the best and hope things don’t turn out as bad as they look at the moment…
The USA is really turning into a dystopia at the moment. Unfortunately, with the elections in Germany just around the corner, I am scared that it will go the same way here. The SPD/Green alliance here could have worked, but it was crazy to bring the FDP into it as well there was no way that was ever going to work out well.
I just hope the AfD don’t gain too much sway. Being an immigrant, even when I am married to a German and I have German citizenship, I still feel unwell, when I look at what is going on, especially in the East of the country.
But the Big Tech and Tech Bro culture seems to be drowning out other bad news at the moment and I agree with you that sometimes Leo goes a bit overboard in praising Big Tech and how AI is the future.
Maybe it is, maybe it isn’t, but what we currently have is certainly not the future. It is too accident prone and fallible at the moment and the problems with data privacy, copyright theft etc. mean that it is too dangerous at the moment.
On the one hand, what Jeff Jarvis and Leo argue is correct, the AI needs as much information as possible to learn from. But on the other hand, they shouldn’t be able to just willy nilly gobble up all copyrighted work without having to license it. Yes, I like things like Perplexity, but on the other hand, if I just use AI to get all my answers, who is going to be paid in the future to generate the information the AI needs to answer those questions?
Likewise, by giving it free reign over the Internet with little or no filtering out of fake news, propaganda, illinformed opinion etc. how is it ever going to be able to provide reliable answers?
These AI services need to be able to provide reliable answers 99.99% of the time for them to be useful. At the moment it feels like we are maybe at 75%, which means you are either intelligent enough to see when it is providing false answers (or to go and check that the answers are correct) or you are going to be misinformed much of the time…
4 Likes
I really think Microsoft gets a bad rap from the tech pundits. The biggest botch about Recall was that it was Opt-Out vs. Opt-In. Which they corrected. Almost everything else was fearmongering by security researchers who were not testing the product under release conditions - with all of the security features in place. The best analogy I saw was that it was like the security researchers got a copy of a bank vault. Rebuilt the bank vault using cardboard. Cut a hole in the bank vault with a knife - and then complained about the poor security of the bank vault. But no one seems to want to condemn them - they want to bash Microsoft.
I think the technologists of the day are predisposed to look for problems with Microsoft. And when you look for problems - you find them. Whether or not they really exist. No, they are not perfect. But they are a lot better than it seems like they are given credit for.
Let’s take a look at Apple’s implementation of its AI - where they completely rewrite news summaries to be wrong. That was enabled by default as well - and you had to Opt Out of that. But they seem to get the benefit of the doubt. Because Apple is magical. If Copilot had a similar feature, with the same problems, there would be no end to the bashing and complaining. But since it’s Apple - well, that’s ok then. I would really like to see all of the tech companies subjected to the same level of scrutiny and expectations.
I use Copilot daily. Sometimes it’s for rewriting an email because I’m struggling with words. Sometimes it’s for coding issues. Increasingly, it’s for more general search functionality. But I use it. Do I love it? Not especially. But I also have no particular complaints about it. I feel it’s better than Google - simply because I don’t get a half-page (or more) of paid ad-driven results. It generally gives me good results.
I’m not saying that there aren’t valid complaints. But it seems to be they are taken out of context and inflated. I expect that on click-driven websites. I really shouldn’t expect it on the flagship tech podcast.
I don’t consider myself a Microsoft partisan as such. But it really frustrates me when it seems like one company is getting criticized for the things that the other companies seem to get a pass for.
The crazy thing is that opt-out as the default is still a thing. So many countries mandate that everything has to be opt-in, it is a wonder that companies either ignore the law or make two systems.
I agree with the fearmongering around Recall - researchers put it on VMs not running Windows Hello ESS (how could they) and had to circumvent all the security protections in order to get it to run, then claimed it was insecure! 
Microsoft has so many genuine problems at the moment, especially with enshitification, that the tech press doesn’t need to resort to such FUD, they could just do some real research and they’d have enough genuine stories without ruining their reputations with FUD.
1 Like
I will agree on the enshittification aspect of things. But they’re not alone. Google results have been enshittified since the late 2010s, and while I do hear complaining about it - I don’t really see anyone changing their behavior.
I think it might have been you that came up with the cardboard metaphor. If so - it continues to be my reference point for the absurdity surrounding Recall. Absolutely brilliant.
1 Like
Yes, I used the cardboard metaphor.
1 Like
Thanks to Ed for the tipoff on the O365 subscription going up. Mine has gone from £79.99 to £104.99 a year with zero notification from MS. I now have 60 AI credits a month, whatever that means.
I might try his trick of unsubscribing and switching to O365 Classic.
I do use Copilot quite a bit, but not in the apps. I use it in areas I know quite a bit about, so am very careful to validate, but it is very powerful in displaying complex datasets in multiple ways without having to do it all manually in Excel.
Update…Ed is correct. I’ve now switched back to my £79.99 subscription, but as I only renewed a few weeks ago (at the cheaper price), I have a year to try out the AI stuff.
1 Like
I’ve been using Perplexity for a lot of search requests on my phone; they were offering a free one-month trial. I thought “no way would I ever pay $20 for this” but after using it for a month I’m seriously considering keeping it. I wish it was more like $5 or even $9 and I had more assurances about privacy.
3 Likes
The problem is, I rarely use a search engine these days, probably 3-4 times a week, unless I am looking up an answer here to help someone. That means I’ve user Perplexity, maybe, 5 times since I installed it at the beginning of the month. The answers are great, but at that volume not worth even $1…
At work, I use search a little more often, but we have the AI ban in force.
Have you tried the free versions of ChatGPT/Copilot and compared it to Perplexity? I was similar, never thought I’d have a use for AI but started to use it just because there’s a Copilot key now on my laptop. I’d miss it now.
I started with Copilot last year and then ChatGPT but I kept bumping up against the lack of more recent information. At work we have access to Copilot and it has been fine using it for background research when I don’t need the most recent information. Unfortunately, we have been limited to Copilot in the sidebar which doesn’t have access to our SharePoint files for deep research into our proprietary stuff, nor is it in Office apps directly. So the main use case has been limited to what you can do asking it to digest a <1MB pdf file or asking about built in general knowledge.
1 Like
it is wild to me how little you use search.
1 Like
I read a lot and the facts and figures stick in my brain and I don’t have to look many things up.
Usually it is just to double check that what I remember is correct, or to prove to my wife that what I said was true.
At work it is a bit different, that is usually looking up error codes or specific PowerShell commands.
At the weekend, we had finally signed over our old house and the double costs of old place and new place fell away so we were recalculating the monthly costs, but she got her payslip and there was less than before. I said it was due to the new health insurance rises from January 1st, she didn’t believe me, so I got Perplexity to look it up, which confirmed what I had remembered.
I’ve noticed that. If you ask something like how many cars are sold every year in the USA, it gives you 2023 stats. But if you then say what about 2024, it has that data too 
From a different source though. Maybe it ranks the older data higher for some reason.
But don’t you ever just wonder which English football grounds are located in a walkable area in their city? Or what is the history of boxty? Or when is sargassum better/worse in the Caribbean? Or where the show The Penguin was filmed? Or which were the most/least successful businesses to have appeared on Shark Tank? Or what the differences are between cotija cheese and queso fresco? Or why do so many businesspeople inappropriately use the term “flywheel”? Or how many CPG products come to market each year? Maybe I just lack the focus to not let my brain go off on tangents 
I’ve seen that before where it will give you something a bit more recent, but I’ve also gotten a response to the effect that a given model doesn’t have more recent information than what I’ve asked about.
1 Like
Not really. Films and stars don’t really interest me. It is rare that I will watch a show or film and want to look it up. There is sometimes and actor who is in a show and I think, hmm, I’ve seen them in something before and look it up, but that is once every couple of months,
All the business people I know speak German and don’t use the term flywheel.
Cheese? I ask the expert at the cheese counter at the supermarket…
For most things, I just go straight to a relevant site.