Profile avatar
alexander.hudek.org
Advisor to Zuva and Vigilant AI, author of AI for Lawyers. Previously I co-founded Kira Systems and served as CTO. I've loved science my whole life and obtained Ph.D and M.Math degrees in Computer Science from the University of Waterloo.
62 posts 58 followers 88 following
Active Commenter
comment in response to post
I don’t actually believe at AI race will be about electrical power. Our current techniques are likely hugely inefficient both in power needs and data needs. The race will be about human expertise in the field.
comment in response to post
He already had access to Twitter data from X. Now X Ai owns it instead of licenses it I guess. But if I were to pay huge sums for AI data I’d ask Elsevier. Twitter data these days would be more like spam than worth money for AI.
comment in response to post
Yikes, thanks for posting this. Toronto needs more community support for electronic, and also other types of music!
comment in response to post
It definitely depends on the lab and the area of study. There are plenty of examples of results that are cited for decades before it's discovered that the original paper is not reproducible. I've also encountered it a lot in applied computer science, where you also have issues of source code use.
comment in response to post
People rarely reproduce results as the incentives are stacked against doing so. Peer review filters very low quality papers, which makes doing related work searches feasible. Especially so in an era of politicized fake science and AI generated papers.
comment in response to post
But put your house in Manhattan and it would indeed be a mansion! It’s all relative. 🤓
comment in response to post
So let me get this right. Poilievre is criticizing Carney for his investments in Brookfield, and right in this very article it says that Poilievre’s investments include…. Brookfield.
comment in response to post
I don’t understand how the dems were threatening crypto. All I found searching is a rule for tax reporting, which seems reasonable?
comment in response to post
I guess a lot of us who follow are people with subscriptions. Definitely a thing though with all the major newspapers.
comment in response to post
And why does Ukraine need to involve the US? They get nothing in exchange for the minerals, and must be forced to publicly grovel for no reason, and EU and others are the ones providing peacekeeping support, what exactly is the US contributing?
comment in response to post
I say a bunch of the blue States should be provinces of Canada. The traditional freedoms of the US don't exist under Trump, and the US is no longer a viable democracy.
comment in response to post
A bit disturbing that it has the answer but doesn't really know why it has the answer and can't figure it out if it's really the right answer.
comment in response to post
Do all your subscribers really want email? I subscribe to one Substack but don’t want email. But Substack makes it nearly impossible not to get email. I only read via the web, never email.
comment in response to post
Armed war if it comes to it. Canadians will die before being taken over.
comment in response to post
Permanently means forever, which is hard to take seriously. Hurt our economy for a long time, sure, but not forever. There are other trading partners, and even the current state of the US won't be forever.
comment in response to post
No, he won’t be deciding the fate of foreigners in foreign countries.
comment in response to post
Because tv advertising is bad for people. There are good ways to advertise but generally forcing your ads on people is not it.
comment in response to post
I find it very hard to believe that “very few” people play with LLMs to explore. They seem very popular. I disagreed with Cuban about his overselling of LLM capabilities. I’m generally pretty pro AI and think it can be positive for society. But it’s important to recognize the limitations.
comment in response to post
This was probably just because of the weather. It was a huge storm that shut down most of the city. Canada has not defunded our air traffic controllers.
comment in response to post
I can’t agree with this. There is still a vast gulf between LLMs and actual domain experts. They are at best at the level of a junior, but not even a good junior!
comment in response to post
So they will attack protestors.
comment in response to post
We need a stronger response. I’d do escalating tariffs every time Trump talks about annexing us. Then if it keeps happening expel the US ambassador and station troops on the border. Threats of war need need consequences.
comment in response to post
Still better than Google. In Canada it is correct. Google put the wrong name in brackets after the right name here.
comment in response to post
For AI, I've switched to mistral.ai. It doesn't get as much attention in North America, but their frontier models are truly excellent, including their code assistant model Codestral that's a part of their API offerings. 4/4
comment in response to post
For CDN, I'm now using bunny.net. It also has Terraform support and must be the easiest CDN + Object Storage + DNS + LetsEncrypt system I've ever setup! Great prices too. The only negative is that their object store doesn't have S3 support yet. 3/
comment in response to post
I've used OVH in the past, and it's come a long way since 2014. In many respects it's a lot less expensive, but it doesn't have as generous free tiers. Also worth considering is www.scaleway.com. I haven't tried them, but they power Mistral, the French AI powerhouse. 2/
comment in response to post
And here is Apple Maps from Canada. This is correct.
comment in response to post
This is what it is from Canada. The brackets should not be there.
comment in response to post
I never get tired of seeing these. We landed on a freaking COMET! ❤️🤩
comment in response to post
I stopped subscribing when their editorial board started taking open right wing positions on the Canadian alt-right "truck" protest in Ottawa. The heavy bias in their writing on the topic was awful. So much for fair and balanced.
comment in response to post
I prefer thought experiments that are based in possible futures. This is not so much a thought experiment as it is a fantasy.
comment in response to post
I agree with this so strongly!
comment in response to post
Not in this thread, but I’ve seen a lot of comments saying things like “anyone who owns a Tesla is a Nazi” on Bluesky in general.
comment in response to post
It's important to remember that A LOT of people bought Teslas or solar panels for environmental reasons, before we knew he was a white supremacist. It's unfair to demonize them or expect them to sell cars, solar panels, at a huge loss, esp. with a recession coming. Don't buy more though! I won't.
comment in response to post
How about we don’t have another military base from a country threatening to annex us on our soil.
comment in response to post
There are plenty of open western ones. The main difference is that often the largest models are held back for a while. But look at models from Mistral, Meta (llama), and Microsoft (phi). These are all useful and good.
comment in response to post
Maybe Canada can ship them corn instead of to the US.
comment in response to post
Having access to computing power does not put you in a dominant position. Advancing AI also requires researchers, and the US is increasingly hostile to those.
comment in response to post
Always neighbours never neigbors.
comment in response to post
Yes! This is a great idea and one that I think LLMs may be able to help with. Few authors fully trace back all references, let alone check them for accuracy. It's a monumental task. While LLMs may not be very intelligent, they don't need to sleep and can read vast amounts of information.
comment in response to post
Questioner? So you’re saying the Daily Show is journalism? I subscribe to the Globe but I sure hope you aren’t implying you’re also a comedy show here. You sure look like one.
comment in response to post
I think every scientist has known the truth of this for most of their lives. I don’t think I’ve done enough to earn the title scientist but I’ve felt it since my PhD.
comment in response to post
This is not true at all.The company I built is just one example (our tech was not based on gen-AI). Speech recognition, translation, even traditional Google search are all based on AI too. It’s frustrating that the discourse has been reduced to lumping everything in with poor uses of LLMs.
comment in response to post
I think it ridiculous that they only get market value. This sort of thing should be done very rarely, and given that, home owners should get double market value at least. It would both soften the blow to homeowners and also make the government think more seriously about what they really need.
comment in response to post
Not everyone here hates AI. <3
comment in response to post
Trudeau promising vote reform and then not doing is still my number one dislike about his party. Frankly, a coalition government that is forced to compromise like we've had is probably the best we can do right now. When a country is deeply divided it feels wrong to give any party a majority.