If getting people to review code is that hard that seems like a problem for our new AI age. AI coding appears to rely on getting people to review a lot code and assumes those people will catch the errors.
From my view a lot of the problems of current AI is that people assume others will review and catch any issues. The manual work is getting pushed around like a hot potato.
> If getting people to review code is that hard that seems like a problem for our new AI age. AI coding appears to rely on getting people to review a lot code and assumes those people will catch the errors.
This is precisely what many of us have been repeating incessantly, for ages now. AI is boosting the wrong part of the ecosystem. It only makes the existent bottlenecks more painful.
Many maintainers get exhausted by code reviews, and recharge themselves through coding. The latter is what AI takes away. It demotes engineers to "AI slop reviewers". I don't know why anybody would sign up for that; IMO it leads to immediate burnout.
China is a mixed economy with some capitalist parts and some socialist parts just like us. Their mix is just a bit more effective than our mix than our mix and they have higher scale.
It's more effective at depressing wages and at shovelling other people's money at whoever the politicians want to win. They are also much better at hiding debt -- in manufacturing companies, in banks, and in provincial governments. A lot of their successes lose money but they are awesome at hiding it and they might well outcompete Western companies and thereby cause a lot of harm.
What do you mean “hiding it”? Are you suggesting Chinas manufacturing capacity is entirely fraudulent or cooking the books? Or that the state is providing subsidies? Because if its the latter… have you seen the brouhaha over Amazon HQ2? Or seen the number of tax credits/incentives doled out by US cities to companies that “promise” jobs but don’t even deliver them? (but keep their subsidies).
China is capitalist on a state level, that's where they are winning. The US lets things get mired in red tape and special interests because nobody wants to take responsibility for growth.
In China, I imagine that if your company does something relevant to the five year initiative then you get a lot of red tape cut for you.
> China is capitalist on a state level, that's where they are winning. The US lets things get mired in red tape and special interests because nobody wants to take responsibility for growth.
i.e. in China, the government controls capital; in the US, capital controls the government.
> If they don't have enough details to implement a solution, they will ask the client for details.
but LLMs are chat bots. Surely eventually someone will set up an LLM based coding system that can ask architecture and design questions before starting to code.
People, including the author of this article, say that design and architecture are the hard parts, but I think long term those are just as solvable as coding.
I think architecture will become like an installer. Some kind of agent orchestration system will ask you "do you want this or that" and guide you through various architecture choices when you set up a project, or when those choices arise.
And for design, now that code is fast and easy to generate, an agent system can just generate two, three or four versions of the UX for each feature and ask "do you like this one, this one or that one?".
So a switch from upfront design / architecture choices you have to put into prompts to the agent orchestration system asking you to make a choice when the choice becomes relevant.
"median American voter" implies a distribution of views like a normal distribution, with a lot of people in the middle and a few people on extremes. If that is the distribution, then the median is representative of most people. I am not sure that is really a great way of thinking about American voters these days. It seems to me that American's views on many issues are tending to cluster around extremes, with fewer people in the middle. So I am not sure the median is as meaningful.
Median does not assume anything about the distribution which is precisely why I use it. Median allows for us to count max total of one category because the variances are so small. Hence why medians can actually demonstrate the underlying distribution instead of commingling amplitude like the mean.
In this case it’s “American Voter” as the category. This is what messes most people up, because they read “American Citizen” but I’m describing only the subset of citizens who successfully vote.
Using that number you’ll see what the demographics demonstrate: there are not as many progressive voters as there are “conservative” voters and only 2/3 of eligible voters even cared to vote.
If you zoom out even further and you evaluate which candidates run, then it really does not matter who is voting or not because ultimately who is on the ballot is dictated by a small group of party leaders, who in turn are dictated by whomever has the most money for ad spending.
I would not be shocked to learn that intelligence agencies are using AI tools to hack back into AI companies that make those tools to figure out how to create their own copycat AI.
My guess is that in Ukraine the Russian EW systems are deployed tens of kilometers back from the line of contact to protect them from artillery strikes and fiber optic drones. These Russian EW systems are likely used to protect command posts and logistics bases but not the line of contact.
But because Iran is not yet an active war zone the Iranians can deploy those systems close to the cities.
Also, Starlink terminals can be located via their RF emissions. So using a Starlink terminal in Iran seems to come with a high risk that security forces can locate and arrest you.
> Also, Starlink terminals can be located via their RF emissions.
Starlink terminals use highly-directional antennas that point at the sky (see. beamforming) and therefore they don't leak much in terms of RF emissions. So unless you can afford to maintain a host of overhead drones on permanent rotation and wide-area coverage, it would be very hard to actually locate anybody. Not that it's impossible, but largely intractable at scale. We use Starlink a lot in Ukraine, and even though the russians have platforms with sophisticated signal processing capabilities (think Xilinx RFSoC) perfectly capable of locating emissions from most communication equipment, they are still unable to locate Starlink terminals. And this is along the frontline, mind you. To cover all of Iran would surely be prohibitive.
I think we will find out that certain languages, frameworks and libraries are easier for AI to get all the way correct. We may even have to design new languages, frameworks and libraries to realize the full promise of AI. But as the ecosystem around AI evolves I think these issues will be solved.
