Okay, so we have a benchmark where the bog-standard human being scores 94.9%.
Then in February (that's three months ago), Microsoft reports hitting 95.06%; the first score to edge the humans.
Then in March, Google notches 95.18%.
Now it's May, and Baidu puts up a 95.42%.
Meh. Swinging dicks with big iron are twiddling with their algorithms to squeeze out incremental, marginal improvements on an arbitrary task.
“Our company is now leading the race in computer intelligence,” said Ren Wu, a Baidu scientist working on the project.
I presume that next month it will be IBM boasting about "leading the race" and being "much greater than their competitors". The month after that it will be Microsoft's turn again. Google will be back on top in August or so...unless, of course, some other benchmark starts getting some press.
I took a look at the actual indictment. Well, at least the first few pages. Remember how people still insist to this day that Bill Clinton wasn't impeached (he was - impeaching does not mean convicting) or that he was impeached for "cheating on his wife"? Years later, the lies spun by his spin doctors still hold fast in many minds. Clinton was impeached for committing perjury in a civil trial. Now the event he committed perjury about was cheating on Hilary, but he was impeached for lying about it while under oath, not for the actual act of cheating on her. Similarly, this indictment isn't really and truly about beating lie detector tests. The government's contention is that Williams had a business whose purpose was to enable people ineligible for certain government jobs to get those jobs through lying and deception. This is defrauding the US government because salaries would be paid to those ineligible people. The government also contends that he enriched himself (through fees he charged) by encouraging people to lie to and deceive the federal government into hiring ineligible people for jobs. The first 6 or so pages I looked at don't actually mention anything about lie detector tests.
One should be very wary of the distinction between "run without refueling" and "run without regular maintenance". Even assuming that the reactor's fuel would last, the ancillary equipment associated with the reactor's operation (coolant pumps and such) and electricity generation (steam turbines) certainly wouldn't be expected to operate unattended and unmaintained for months, let alone years.
That said, the fifty-year planned lifespan of the Nimitz-class includes, if I'm not mistaken, a mid-life refuelling and complex overhaul (RCOH). To be fair, the reactor's fuel would likely last longer than the planned 20-25 years if the carrier weren't actively steaming--but I wouldn't trust the other parts to last anywhere near so long.
After dialup disappeared, AOL had plenty of cash in the bank. So they became a type of venture capital. They bought Huffington Post, Tech Crunch and many others. Since they actually have a lot of web traffic, they started an advertising business.
Thank you for this explanation. I was really struggling to understand why Verizon would want to pay so much for the dial up business but clearly they want everything else and are just taking the dial up business as part of a complete package, not specifically trying to get that.
The solution: leave the phone at work when you are off duty.
That would work...except that the employer insisted that she keep the phone with her and powered on at all times. According to the claim, she was on call for client emergencies, even when off the clock.
... the real question is, "Were the accidents something a human driver could have avoided?"
It's an interesting question. On the other hand, most collisions are something a human driver could have avoided somehow...but didn't.
Sometimes you have to yield right-of-way because it's clear the other driver isn't going to. Do autonomous cars know that?
I would be shocked if they didn't "know" something like it. I can't imagine any car (let alone the entire group of 44 which didn't have a collision) doing a full year of city driving without encountering multiple situations where another driver failed to appropriately yield the right of way.
As a New Yorker, I much prefer LaGuardia, and strongly disagree with calls for its closing.
The point is, I think, that in exchange for an improvement (real or hypothetical) in convenience for a small fraction of total air travellers, there is a substantial and arguably unnecessary burden of cost and inconvenience to the entire system (which is ultimately paid for out of everyone's pockets--and user experiences).
I would love to see these large airports replaced with multiple smaller airports. A larger percentage of the population would have an airport nearby, and average travel times would be reduced significantly.
Well no, it wouldn't. A fully-served point-to-point network with n nodes (cities served) has on the order of n squared links between nodes. The number of passengers desiring each direct link gets to be very small, very quick, meaning infrequent scheduled flights on small, underfilled, costly-per-seat aircraft. So what happens is that airlines adopt (to one extent or another) a hub-and-spoke model. Most direct point-to-point routings are dropped. If I want to fly from East Podunk, NY (POD) to Los Angeles, I can't get a direct flight POD-LAX. Instead, I get a hop to an airline's hub (JFK or ORD or DTW or wherever), and a connection from that hub to LA: POD-JFK-LAX, or POD-DTW-LAX, or POD-ORD-LAX.
If I want to go to a destination served by a smaller airport (let's call it West Lemon, CA: LEM), then I'm taking three flights: spoke to hub, hub to hub, hub to spoke: POD-JFK-LAX-LEM. And each of those flights carries with it the time penalties associated with loading and unloading passengers and cargo, and a risk of delays or cancellations due to weather and other circumstances--plus the plain old waiting for connections, because service to and from the small airports at POD and LEM is infrequent.
Worse still, all those little commuter flights linking the regional airports to the major hubs are going to take up gates and takeoff and landing slots at those busy airports, slowing down the whole system and/or pushing those less-important flights to less-desirable times of day. Taken all together, offering frequent (or even just daily) service to a lot of small airports is going to mean a lot more flights of a lot more smaller aircraft, and/or passengers frequently making multiple connections. It would be expensive per-seat and vulnerable to failures and delays.
Now, La Guardia is an interesting case. Since it's right next to downtown New York, it draws a substantial number of departing or arriving passengers, and enjoys a kind-of-weird pseudo-hub status for historical reasons. Practically speaking, though, it means that there are effectively two hubs (LGA and JFK) or even three (if we count EWR) serving the same area, resulting in needless duplication of services. Routes that could enjoy frequent service with inexpensive (per-seat) full-sized jets get less-full or more-expensive aircraft, or less-frequent services divided between two or three New York destinations. Local New Yorkers enjoy the appearance of convenient, direct flights, at the cost of making the rest of the system a bit worse and a bit more expensive for everyone.
police injury rates are _much_ higher than most work
Welp...sort of. The U.S. BLP recently published their 2013 census of fatal occupational injuries. The overall fatality rate for the workforce was 3.3 fatal injuries for every 100,000 full-time-equivalent workers per year. Management employees averaged 2.4; sales 1.6--no surprises there, really.
For employees in the "protective service occupations" - police, firefighters, correctional services, animal control, security guards, and so forth - the rate was 6.9 fatalities per 100,000 FTE. (I haven't been able to find data broken out by occupation within the category. If someone can find that, that would be great.) So that's what we expect--police, firefighters, and others do have a riskier job than the average, and riskier than the typical office worker. Somewhat surprisingly, the relative risk is only a factor of three or four different when comparing a police officer to, say, an IT manager.
But...there's the rest of the table. "Intallation, maintenance, and repair" occupations? 7.2 fatalities per 100,000. "Construction and extraction"? 12.2. "Transportation and material moving"? 14.9. "Farming, fishing, and forestry"? 23.9.
The real manly men, in real danger on the job, are apparently out there working with tools, building stuff, drilling for oil, driving big rigs, and cutting down trees.
And let's be honest--a lot of the injuries and fatalities sustained by police officers aren't directly attributable to violent suspects. A big chunk of them come from the fact that the typical frontline officer spends a lot of time moving around--in a patrol car, on a motorcycle, on foot, or on a bicycle. Special laws protecting police officers from insults don't actually reduce their likelihood of being in a vehicular accident, or getting clipped by a passing car during a traffic stop, or slipping on an icy sidewalk in the winter. Looking at the last ten years' police fatalities for the United States, the total number of officers killed in motor vehicle incidents (car and motorcycle crashes; hit by car) is 605. The total number of officers fatally shot, strangled, or stabbed is 553. (And I suspect that the proportion who get shot is even lower in Canada.)
The elaborate charade is all about convincing Congress that the negotiation is so complex that the president NEEDS fast-track authority to get this whole deal done.
Well, Bush asked for this kind of authority too, so do note that this not particular to Obama. The real reason the president wants this is to prevent individuals from tagging on bill busting riders where the president would have to veto his the agreement to stop some unacceptable after the negotiation condition from taking place which is exactly what the person wants who tagged the rider onto the bill. I never hear about other countries having this kind of problem. Can you imagine if you agreed to buy a house at a certain price and then you show up for closing and the owner says "Surprise! I never told you this before, but you have to buy me a new BMW to get the house." Nobody would go for that. But doing similar things in legislation is completely OK apparently. If you don't understand why all presidents regardless of party affiliation can't trust Congress to just leave the agreements alone before voting on them, then you don't understand why this is necessary.
even though the shuttle didn't have an equivalent system for many conditions - see challenger
That was true for 1986 NASA, certainly. Post-Challenger there were major changes (extensions) to the list of abort options - including a new bail-out capability - which made the loss of two engines crew-survivable for the entire ascent, and the loss of all three main engines survivable for most of the ascent. (See the Wikipedia article for details.) As it turns out, we have higher expectations 30 years on. Whodathunkit?
A commercial plane will most probably undergo through several maintenance events and checks during that sort of time frame, where cycling the power is part of the procedure.
It's very reassuring to know that it probably won't happen.
As other posters have noted, 248 days of operation means skipping twenty-plus maintenance and inspection cycles, plus missing one or more engine overhauls. That sucker's going to fall out of the sky due to a hardware problem before the software error gets the chance.
Even in the absence of regular, scheduled, required maintenance, there will be hardware failures due to stuff wearing out, with sufficient frequency to force reboots at less-than-eight-month intervals. Honestly, the FAA is going to ground any airline that was so lax as to get within six months of tripping over this bug.
That's not to say that this bug is a good or acceptable thing, nor that something like it couldn't have much more serious effects. But this particular error is a non-issue from a real-life consequences standpoint.
I don't know about the technology or the algorithm(s), but the linked article is certainly nonsense.
“You could use the same basic framework to do robust decision making like trying to come up with insulin and glucose monitoring plans [for diabetes patients],” says Neil Burch, a computer scientist at the University of Alberta who helped design a poker-playing AI earlier this year. “You get regular snapshots of glucose levels, and you have to decide how much insulin you should take, and how often.”
Look, I get it. Nobody wants to admit that they're spending their grant money this way because it's fun to get a computer to play Hold 'em. But that's got to be the dumbest justification I've ever read. Human metabolism is complex, but the pancreas doesn't bluff.
This is basically a beginning poker player (fresh blood) but who is more consistent. A pro will absolutely clobber it.
In other words, either the researchers involved are complete idiots, or a Slashdot poster jumped to a useless conclusion based on a strawman argument spun from the summary. Hm.