But that won't protect me from accidents caused by algorithm failure by other auto-braking cars
If you have an accident because the car in front of you makes an emergency braking, blame is not on the algorithm. The safety distance must be such that at any time the car in front can brake and you're able to stop behind it at your current speed.
n/t
are you stupid or what? just because we don't know how our machine between our ears work, does not mean it's not a machine. Championing for a supernatural cause this way, is weak, isn't it? Religion is not to think that we are machines, religion is to believe we are not.
A brain is not a machine any more than a lizard is a machine.
There's no supernatural implications in differentiating between a mechanism built for one purpose, using pieces that were put there to serve that purpose, and a whole system that has evolved and is able to function by sheer adaptation to pressures that destroyed any similar system that didn't work that way.
The two entities belong to different categories. And calling the second "a machine", merely because it's also physical, doesn't work beyond a metaphor and is harmful to understanding.
It is not, and never has been, a judges responsibility to present evidence in court.
That is the job of the lawyers representing the defendant. It is also their job to request and show reason for further investigation.
And that's the precise reason why the legal system in Common Law countries is utterly fucked-up and it's impossible for the poor to get actual justice, except by chance. You only get the evidence and legal arguments you can pay for.
In Roman law systems, it is absolutely the responsibility of the judge to ensure that enough evidence is collected to properly assess the case and provide an accurate judicial truth.
Now don't get me wrong, the poor can also be screwed up in a Roman law system, in special in case of corrupt cops or public officers. However, they can still get a chance if they get a honest judge set to discover what actually happened, and who follows proper procedure.
> and what seems natural would probably look at a lot different.
It would not. It's in the nature of open unregulated competition systems that any random initial advantage produces a power imbalance that accumulates more and more resources in fewer hands. "Money attracts money". The only way to fight back in such systems is survival of the fittest, accumulating enough of your own and extinguishing those who can't cope.
It takes an active constant redistribution effort to take a part of the accumulated resources and return them in equal parts to those left behind. That's why taxes exist.
Some engineers have praised LabView and clones for such.
Yes, I have worked professionally with arrow-and-boxes languages and I enjoy their data-flow oriented workflows.
However in my experience I find that these have at least two weak points for anything more complex than ETL / Data warehousing:
- large programs tend to be hard to read, either they get too many crossing wires or too many boxes-within-boxes
- control loops, specially nested loops, are hard to design. Missing or hard-to-use recursion, combined with the need to explicitly connect every data dependency (that in written code would be solved with separate declare / assign statements) makes it hard to think in high-level control structures.
Also text language have the advantage of a linear reading flow; what is a disadvantage for designing a program is an advantage for reading other people's code.
Same team here. I completed Computer Engineering, and can visualise the structure of quite complex systems as long as I can figure out their connections as spatial relations, up to very abstract levels.
But in terms of coding, I'm quite slow with text programming languages. I need to translate the spatial relations in my head into a linear sequence of symbols and words, losing most of the structure, and remembering the exact syntax needed at each step is hard.
I dig every visual feature in IDEs that brings back some of that structure, I love indentation-significant languages, and I usually like the Visual languages that other programmes dislike; though these languages usually have a very low abstraction ceiling.
To be an efficient coder I would need a high-level general-purpose visual language, and those are usually just modeling languages, they don't generate proper executable code. The closest thing I use are spreadsheets and web notebooks, where I can organise the global structure of the individual items in the program, without the exact language syntax being an essential component.
Go ahead, read that Bible of yours and watch the progression from "God is one of the gods" to "God is the only god" over the course of the text.
The godcides in between were thoroughly hidden, all traces covered.
Digital toxic girlfriend
No, the GP is correct. Psychologically no one attaches as much importance to their own mistakes as to those of others, so your "culture of assuming your personal responsibility" cannot be the sole cause of the system working better; if it were, a culture of "uncovering the mistakes of colleagues and making them pay for them" would work even better.
What really improves a system is "making sure that mistakes are discovered and corrected, no matter who makes them". Owning the errors in your work may be a part of that, but the essence is that a culture of finding a culprit and blaming them is not good, what matters is what you do to fix it afterwards. Blaming individuals is as bad as blaming the system if it's used merely as an excuse for not solving the problem.
It's not as if balance accounts are physical magnitudes. Accounting is a human tool used by humans to make decisions, a number that summarizes a million inputs into a comprehensible codified simulation.
In a system as big and complex as a country's finances, what matters most for making decisions is the changes in the inputs, that causes the value to be updated (cybernetics 101); so, the actual exact value only matters to the few companies or projects on the margin, that would be placed in a different class in the decision process if the value had been slightly different. It sucks to be them, but unless the system is on the bringe of a phase transition, it doesn't affect the stability of the whole thing. It's likely that the impact of discovering the mistake can have a bigger impact that the mistake being there to begin with.
Reality has a well-known racial bias.
Given that race doesn't exist, that claim looks dubious.
I can't see how a mechanical turk activity could be done at the speed that it's being done by chatgpt.
Not in all cases, but in many of them you could:
1. Do a search query from the content
2. Show X search results to the Mechanical Turk worker
3. The worker chooses the most correct search result for the given the query.
4. The text in the result is then presented to the user.
Not all mechanical turk tasks need to be about generating new content. Simply giving an answer with the confidence of a human in the loop supervising the result could improve AI results for many tasks.
So, you consider the coffins, caskets, and headstones in your local cemetery to be contaminating "garbage." Interesting.
They certainly have an interesting mixture of micro and macroscopic fauna.
Why risk spreading bacterial life on a body that does not have it, with actions without a direct benefit?
The flow chart is a most thoroughly oversold piece of program documentation. -- Frederick Brooks, "The Mythical Man Month"
