AI in 2025

The "arms race" for who will build the best AI is still dominating the news in the new years 2025. 

An interesting article I read discussed, among other things, the new Google Gemini:

A nice example of that is Google's Gemini with Deep Research (accessible to everyone who subscribes to Gemini), which is really a specialized research agent. I gave it a topic like "research a comparison of ways of funding startup companies, from the perspective of founders, for high growth ventures." And the agentic system came up with a plan, read through 173(!) websites and compiled a report for me with the answer a few minutes later.

The result was a 17 page paper with 118 references! But is it any good? I have taught the introductory entrepreneurship class at Wharton for over a decade, published on the topic, started companies myself, and even wrote a book on entrepreneurship, and I think this is pretty solid. I didn't spot any obvious errors, but you can read it yourself if you would like here. The biggest issue is not accuracy, but that the agent is limited to public non-paywalled websites, and not scholarly or premium publications. It also is a bit shallow and does not make strong arguments in the face of conflicting evidence. So not as good as the best humans, but better than a lot of reports that I see.

Still, this is a genuinely disruptive example of an agent with real value. Researching and report writing is a major task of many jobs.

If you work with your head more then with your hands I guess your job is in some danger in the next 3-7 years. 

Read the whole article here: Prophecies of the Flood. 

How to disable Microsoft365 payments for AI but continue paying for Windows.

I ran into this interesting "Toot" on Mastodon from @timixretroplays@digipres.club today, about Microsoft raising the subscription for Microsoft 365 with about 50% to recuperate some of the costs of their LLM/AI adventures.

There is a way to only pay for windows and not pay for the AI stuff, but it's hidden away under the "Turn off recurring billing" feature. 

I am not very active on Mastodon, but thanks to OpenVibe I am still able to follow Nostr, Mastodon, Bluesky and now Threads (but I don't use that). 

I am noticing the most of the interesting stuff that I find is either on Nostr or Mastodon. Bluesky seems to be more political, which I don't really care about.  I know this is a feature of who I follow and not just the platforms as a whole, but still…

Anyway, I hope this saves you all some money, as, no doubt, 2025 is going to be another expensive year. 

Is Apple ready to deprecate Music.app?

There used to be a time when you could use iTunes to upload your own music collection, and have it ready to play on all your Apple devices, whether the music was purchased through the iTunes Store or not.  Back then, you could also use iTunes to rip a music CD and save the music files in AAC or MP3 format. Nowaday's it's still possible to import your own music in the Music.app on your Mac, but somehow these songs are no-longer synced between devices. So you have to do this on every Mac separately, and also your other iDevices (iWatch, iPhone and iPad) don't see anything of this music at all. 

I weep for it becoming increasingly hard to buy and own your own music. 

All subscriptions for streaming music services are around $10-$20. If you want to listen to new music often, then this is a great deal. If, however, you are like me and want to listen to the same 1000 or so songs often, then this is a horrible deal. Better to buy these songs outright and just pay for them once. 

I bought my first iPhone (model 4) in 2010, and from that moment onwards I started buying music in the iTunes Store that came installed on the phone. That means that I've now had 14 years to buy music at Apple's. I have 4.33 GB of songs downloaded on my iPhone right now, so that is about 866 songs (assuming 5 MB per song), or 5,15 songs bought per month over a 14 year period. If we assume every song was priced at $1,29 (the most expensive price option), then I would have payed $6,64 per month over this 14 year period. That's quite cheaper than the $10,99 per month that Apple is now charging for their unlimited music streaming service. And this is precisely the reason why Apple is doing all it can to undermine the buying and playing of songs on their iDevices.

I don't know of any official statements to the effect that they want to stop the buying and owning of music, but the UX in the Music.app makes this very clear:

• As mentioned above, the Music.app no longer syncs your own imported songs.
• Since a few iOS versions back now, the playlists that you make on one iDevice no longer sync with the other iDevices. More specifically I see all my playlists on my Mac, most of my playlists on my iPhone, none of my playlists on my iPad and most of my playlists on my iWatch (I assume the iWatch sees what the iPhone sees?). 
• To make matters worse, 1/4 off all the UX real estate in the iPhone Music.app is now dedicated to advertisements to push you to the subscription model.
• And, just today, I found out that a song I once "bought" is no longer playing on my devices. The song in question is "Diamond Heart - Single" by Alan Walker and Sophia Somajo. I don't know if is simply a strange bug or that these artists have received their royalty payments from Apple years ago and now have found a way to deprecate music that I payed for from my devices?  

I remember finding on Hacker News some years ago an article by the BBC about the deprecated Microsoft book service. It was called "The e-books stopped working".  I remember bemoaning the fact that a book can "stop working" back then. I always assumed that the Roman Empire fell because of the increase in complexity making it harder and harder to do the simple and sensible thing to do. Is this how all empires fall?

I don't know, but for now I am gloomy on the prospect of us ever having a simple and affordable way to buy, play and own our own music. Maybe this is the reason why I usually buy my books in the dead-tree version and only my pulp holiday reading in ebook format. 

New use case for AI? Let it fact- and math-check your work



I am very interested in what "AI", or better, "LLM", are capable of. It's been two years now since I started using them for translation and research purposes. But I'm still finding that they make all kinds of hard and easy to miss mistakes. Of course, I mostly notice this when I ask questions that I already know the correct answer to. Gell Mann Amnesia is live and kicking, I guess. 

Anyways, to invert the trust model, this blog post suggests an alternative use of LLM, namely to use them to fact check your work. Or in this case, check the math of a science paper. 

As one fun example, I read an article about a recent social media panic - an academic paper suggested that black plastic utensils could poison you because they were partially made with recycled e-waste. A compound called BDE-209 could leach from these utensils at such a high rate, the paper suggested, that it would approach the safe levels of dosage established by the EPA. A lot of people threw away their spatulas, but McGill University's Joe Schwarcz thought this didn't make sense and identified a math error where the authors incorrectly multiplied the dosage of BDE-209 by a factor of 10 on the seventh page of the article - an error missed by the paper's authors and peer reviewers. I was curious if o1 could spot this error. So, from my phone, I pasted in the text of the PDF and typed: "carefully check the math in this paper." That was it. o1 spotted the error immediately (other AI models did not).

Even if it is unwanted or unethical to have an LLM write your paper, you can still make heavy use of it: simply ask it to read along and fact/math check your work. 

My dystopian week

This is a cross-post to our brother/sister platform, so you can either read this post on janromme.com or on seriousaboutech.com. 

Last week the well-known software engineer and investor Marc Andreessen (of Mosaic/Netscape fame) went on a long interview at with Joe Rogan in which he explained, to I imagine the shock of millions of listeners, the existence of a pressure mechanism barely controlled by rule-of law namely: debanking.

This is like deplatforming. When you are kicked off Twitter or Instagram, you are deplatformed. When the bank cancels your bank account and tells you in a letter, phone call or email that you have x days to take your money and leave, that's called debanking. 

David Marcus worked at Meta (Facebook and Instagram are well-known companies of Meta) to set up their cross-border payments platform Diem/Libra, and were targetted by what I would call Law-fare (it's a wordplay on warfare). Many others chimed in to relate similar harrowing stories of how they had to keep cash under a mattress because not only did they lose their bank account, but also would no other bank start a banking relation with them. 

You could be excused for thinking these things only happen to rich and powerful people with similar adversaries somewhere far away in that parody of a country, the United States. 

Let me tell you about my past week now. 

Since we adopted the "Black Friday" craze here, I wrote a comparison of mobile phone subscription plans and set my mind to becoming a customer at an MVNO called Ben. After filling out some form for number porting and such like, I received an email that I am rejected as a customer. The reason is that one or multiple of these 3 companies has deemed my "not creditworthy enough": Preventel, Experian and Focum. 

I always pay my bills and have no outstanding debts anywhere, so this is an obvious mistake. 

But I didn't have the time, energy or even the impulse to challenge these 3 companies (that I never heard of before that day) so I just went to the next telecom provider on that list that was 33 cents more expensive. 

But that was not all…

Some elderly friends of mine, who live down the street, complained about their slow iPads. So I told them I would help them order new iPads. They both wanted the iPads Air M2, which, I think, is a good choice because the iPad A17 is now old tech and the iPad Pro M4 is expensive overkill for almost anybody.  

I had my laptop with me, so I proposed we order both devices and fitting cases from my laptop. 

One day later, one of these ladies is in the supermarket, and she discovers to her horror that her debit card is blocked. I don't know if/how she paid at that instance, but she right away went to her bank to complain. The explanation for her being blocked was that with her bank account, a payment was detected by CrowdStrike to have originated from a "laptop that also does crypto transactions". 

The fact that she had to authenticate and confirm these transactions and the fact that we just ordered something on Amazon, and the fact that this, according to the bank, "suspicious" transaction was not blocked, but the next one was, is mind-boggling to me.

I am happy to confirm that after complaining at the bank, her bank account was unfrozen again. When this happens to you for the first time, this adage: The money in your bank account is not yours, not money and not there rings in your ears. 

How easy would it be for an well-meaning but misguided algorithm, AI or a semi government organization to debank or block whole swats of people that it, or they, deem 'unwanted'? Our societal reliance on these choke points makes us fragile. This is one of the reasons why the Human Rights Foundation and others are so sceptical of Central Bank Digital Currencies, the latest efforts to do away with cash all together. I know my individual actions, trying to pay with cash wherever i can for example, will in no way turn this tide. So why bother trying?  

 

We live in 1984-esk times. End Times. And a worldwide time is coming in which "nobody can buy or sell without having the mark of the beast". So things will get weirder before they get better. 

On that happy note: until next time 👍 

Using Citations to Explore Academic Literature | Inciteful.xyz



Incitefulxyz lets you upload a citation for a paper, and it will point you to connected papers that where not referenced. Basically, it helps you spot the "blind spots" of the papers authors. 

Another way to use Inciteful is to upload two seemingly unconnected papers. It will then find the links between these papers.  I have not had a need for this function yet, but I can imagine it being useful when writing a dissertation. 

Give it a try here: https://inciteful.xyz/

Anyone Can Buy Data Tracking U.S. Soldiers and Spies

Reading posts like these make me wonder, do military personnel not learn to instal add blockers on all their devices? Are they allowed to use devices that spy on them, like an Alexa or Google smart speaker in their living spaces? 

But also, if the US military can't stop a bunch of companies from spying on their personnel, how could you and I hope to defend our privacy? 

Read Nicks take here: 

Anyone Can Buy Data Tracking U.S. Soldiers and Spies – Pixel Envy pxlnv.com