> they’re made for handling language, not playing games with strict rules and strategies
Here's the opposite theory: Language encodes objective reasoning (or at least, it does some of the time). A sufficiently large ANN trained on sufficiently large amounts of text will develop internal mechanisms of reasoning that can be applied to domains outside of language.
Based on what we are currently seeing LLMs do, I'm becoming more and more convinced that this is the correct picture.
Was the ketamine not prescribed to him? That makes it an example of the kinds of abuse/misuse we could expect more of if these things become more widespread, if nothing else.
I'll throw this out here as well: Is there any decent alternative to GitHub Copilot when using Visual Studio? (Pretty happy with it to be fair, but would be open to trying others.)
> What this tells us for AI is that we need something else besides LLMs.
An easy conclusion to jump to but I believe we need to be more careful. Nothing in these findings proves conclusively that non-verbal reasoning mechanism equivalent to humans couldn't evolve in some part of a sufficiently large ANN trained on text and math. Even though verbal and non-verbal reasoning occurs in two distinct parts of the brain, it doesn't mean they're not related.
If you happen to find yourself in a relationship where arguments never happen, then by all means. But as someone married to his own Margaret for the past decade I find many of these hilariously relatable.
Even if she can drive me to the brink sometimes (and vice versa), I still very much love her, and when looking at it soberly the good far outweighs the "bad". Taking a bit of craziness on the chin now and then is a minor nuisance compared to many other problems in this life, and a price I'll happily keep paying for the sake of my family.
If you go further down (to the level of fundamental physics) you'll see there's plenty of mystery left. Even in our technological age, we still don't even have a solid explanation for what makes an apple fall to the ground. The world is far from dull.
Just to confirm, I have strongly nostalgic memories of gaming magazines from the 90s, and recently decided to bring my 8 year old son to the biggest international magazine shop in my northern European city in hopes of giving him a taste of the joy they brought me. But the selection these days is simply garbage. Only a couple of gaming magazines remain and they were very gimmicky with little journalism or substance.
I'd wager if Linux had the same userbase as Windows, you'd see more ransomware attacks on that platform as well. Nothing about Linux is inherently more secure.
As someone who frequents poorer regions of West Africa, that is exactly what I had in mind. Economic growth driven by increased use of hydrocarbons could massively help the populations there.
For me the test is: What unethical action would optimize profits for company X? And what is the control mechanism that prevents them from doing so? If the answer to question 2 is None, there's a good possibility the answer to 1 is already happening.
Their working domain. If you're a developer in financial services, learn finance. If you're a developer in oil exploration, learn seismology. Don't constrict yourself to only being a programmer who writes whatever code you're told to. Be a holistic problem solver.
The risk if climate science is wrong: In the west, we only make a bunch of imaginary problems for ourselves that waste the time and money of the average citizen on a daily basis. But in the global south, billions of people are forced to live in perpetual and worsening poverty with no chance of economic improvement.
If climate science was right and we still allowed carbon to burn, yes the world would heat but if we used our minds I'm sure we could solve a lot of the challenges as they appear.
Well, 'just science' is easy to say when you're not suffering the direct impact of climate politics. Looks more and more to me like they're trying to hold back the economic development of many of the poorer countries in the world. The models they base this on at least better be correct.
Under the impression that a lot of climate scientists are quick to disregard the effect variations in solar activity has on climate. To be clear, I'm not denying the effects of CO2, but it still seems possible that the role of CO2 has been exaggerated to some extent. The rise in global temperatures in the past century has coincided with increases in both CO2 and solar activity. And it's probably not a coincidence that last time we had a long period with a quiet sun (the Maunder minimum) was also a very cold period here on earth.
Will be interesting to see the effect on global temperatures when this SC is finished, especially if we get a prolonged quiet period afterwards. Might provide a proverbial cold shower for climate scientists.
MS isn't starting from scratch, it already has the weights of the worlds most powerful LM, and it's all running on their datacenters. Even without Sam, they just need to keep the current momentum going. Maybe axe ChatGPT and focus solely on Bing/Copilot going forward. It would give me great satisfaction to see the laughing stock search engine of the past decade being the undisputed face of AI over the next.
I taught myself to code in high school and then did the 5 year CS degree at university. School taught me a lot of interesting stuff that I probably would've never picked up on my own, and gave me a much more grounded perspective on the entire field. But very little that was directly relevant for enteprise programming (which I've been doing professionally for over a decade by now).
Even if the physics is well established, I believe you need people who have spent their 10,000 hours learning and practicing with it in order to make the best use of it.
Maybe Microsoft holds some sway? Sam made a snide tweet at them a few days ago. Running the whole Dev Day demo on Mac's might also be taken as some kind of statement.