ggreer
It's not feasible to use Pangram in a feedback loop with an AI. Pangram only reports whether it thinks the text is written by AI or not. It doesn't report whether the text makes sense, or whether it contains errors in spelling or grammar. So when people try to pair AIs with Pangram, the quality of the output tends to degrade to the point where it's not useful.
Basically, it's a harder problem to get an AI to accomplish some goal (such as writing a blog post) while also having its output mimic human text. To use an analogy, it's like building a stealth aircraft with the same speed, maneuverability, range, and payload as a non-stealthy aircraft. Even with a much larger budget, it's not always feasible.
1. https://www.pangram.com/blog/no-llms-dont-just-mimic-human-t...
Both of these examples are nowhere close to the power given to the government before Citizens United: a censor who chooses what people are allowed to read or hear in the months before elections. Unlike copyright or contract law, the punishment was before the fact, not after. And it applied to films people paid to see, not just ads.
You're saying you want Trump's FEC to decide what ads can be shown during a football game. Do you really think that would be an improvement?
If you think voters’ minds are so feeble that they can be swayed by propaganda, then the problem is the voters, not the propaganda. Anyone so easily manipulated can’t be trusted to decide who is in power. It is incoherent to endorse restrictions on free speech while also supporting universal suffrage.
Yes, I was using an analogy.
I don't know what's going on with you today. You get hung up on language even when you know what I mean. You argue from authority (eg: "former gas station owner here", "I use Lithium Ion batteries in volume"). You don't address my stats about vehicle fire fatalities. You don't explain why you think my conclusion doesn't hold up. And you won't say what would change your mind.
We're not having a debate. You're just contradicting me.
Diesel vehicles usually have much larger fuel tanks than gasoline vehicles, so while the chances of a crash causing a diesel fire are lower, the consequences tend to be worse. If the stats are to be believed, the two factors balance each other out.
The relationship between energy stored and energy released was mostly determined by how big the batteries were, not their state of charge. Only 4 of their tests varied state of charge, and those were all in vacuum (with cells #1 & #6), and state of charge was never below 60%.
I did correct for installed base. I'm using fatalities per vehicle mile traveled. As I said, diesel vehicles are 11% of vehicle miles traveled in the US and responsible for 12% of vehicle fire deaths in the US. EVs are 3% of vehicle miles traveled and less than 1% of vehicle fire deaths, so they're safer than diesel. The only way this could be false is if battery fire deaths are grossly underreported. That seems unlikely to me, as deaths caused by any new technology tend to be overreported. It should also be surprising that battery fire deaths are so rare, because we have little experience dealing with them. Over time, we should expect the danger of battery fires to decrease as technology and emergency training improves.
At this point, I don't know what would change your mind on this topic. I would change my mind if stats showed higher fire fatality rates per vehicle miles traveled for EVs, but that doesn't seem to be the case. I feel like I'm arguing with an anti-nuclear person. Statistics show that nuclear is extremely safe, but anti-nuclear activists ignore stats and point to nuclear accidents and the dangers of radiation. Meanwhile they ignore or discount the harms caused by alternatives. I don't see how your line of argument is any different.
Diesel vehicles are 11% of vehicle miles traveled in the US, and around 60 people died in diesel vehicle fires last year (12% of vehicle fire fatalities). So they seem to be as dangerous as gasoline vehicles. This is probably because while diesel is more difficult to ignite than gasoline, diesel vehicles tend to have larger fuel tanks, so the harm caused by a diesel fire tends to be greater.
The fact that some people have died in lithium battery fires doesn't tell you if they're a greater risk than petroleum vehicle fires. Around 500 Americans die in vehicle fires every year. Last year, EVs were around 3% of vehicle miles traveled in the US, so if they're as dangerous as petroleum vehicles, we should expect 15 battery fire deaths in 2025. The actual number seems to be 2 or 3.
Over/under-charged batteries are a safety hazard because they can generate hydrogen gas which can initiate a fire, but the energy released from the fire will be pretty much the same.
Most improvements have been in things like modularity, sighting systems, and standardization for mounting accessories such as lights and laser aiming modules. Better metallurgy does mean that modern firearms tend to last longer than older ones, but even a firearm from WWI can fire thousands of rounds before the bore erodes enough to hurt accuracy. I doubt we'll see any significant change in small arms unless something revolutionary comes along, such as portable energy weapons or mass drivers.
The main causes of incompatibility were the transition from ball-and-powder to rimfire cartridges, then the transition from rimfire to centerfire. It's surprising how little firearms have changed in the past century. 9mm Parabellum is 125 years old. The Browning short recoil action (used by most handguns today) is over a century old.
1. 66,625 vehicle thefts divided by a population of 27,600,000. https://www.abs.gov.au/statistics/people/crime-and-justice/r...
2. 659,880 vehicle thefts divided by a population of 341,784,857 is 193 per 100k. https://www.nicb.org/news/news-releases/us-vehicle-thefts-ex...
If you measure auto theft rate per vehicle (not per person), then the US has the same rate as Canada. There are 16% more auto thefts per capita in the US, but the US also has 17% more cars per capita than Canada.
If lack of immobilizers was the main factor, Canada would have significantly lower auto theft rates than the US, and countries that mandated immobilizers would not have theft rates 15x higher than countries where they were uncommon. Neither of these are true, so immobilizers are most likely not the main factor in auto theft rates.
Auto theft rates are mostly determined by car ownership rates and the number of auto thieves on the street. The number of auto thieves on the street is determined by how quickly they are caught, and how long they are locked up for. Auto theft recidivism rates are higher than any other crime, including drug offenses. Around 80% are re-arrested within 3 years of being released. Considering the clearance rate for auto theft is around 10%, that means the average auto thief is stealing a lot of cars before being caught again, and the actual recidivism rate is higher than 80%.
1. 4.7 per 100k people per year vs 288.5: https://en.wikipedia.org/wiki/Motor_vehicle_theft#Motor_vehi...
Lastly, are you admitting that your first link was incorrect since you didn’t respond to my points at all?
Working backwards from their stats & FARS data, I calculated that for their claims to be true, each Model Y would need to accumulate only 4,300 miles per year. This is far below Tesla's reports of 13,000-14,000 miles per year. Also iSeeCars claimed that an average of 2.8 vehicles were involved in fatal accidents for every billion miles traveled. But at the time the US had 13.3 deaths per billion miles traveled. So for their data to be correct, the average fatal accident would need to have five fatalities.
At some point, I think electric VTOL aircraft will make economic sense. Companies like Vight Aero are working on two seaters with decent range and speed.[1] Then you'd just need an area the size of a tennis court to take off and land. You also wouldn't have to ship fuel in, as you could charge the aircraft with off-grid solar. If that happens, I could see a lot of people moving away from cities.
I bought 20 acres bordering the national forest a couple years ago. The seller made sure I had some idea of what I was getting into. Due to the isolation and terrain, everything is more difficult. It's been a lot of work to build a road, level an area to build on, construct a shed, and set up off-grid solar power. But it's also very rewarding. The views are spectacular, and at night you can see the milky way. There's still a lot more work to do before it's worthy of being called a vacation spot, but I'm very happy with my decision.
1. https://www.zillow.com/homedetails/Snag-Mountain-Rd-110-Wash...
The biggest creator of space debris is China. Around 30% of all ISS debris avoidance maneuvers are to avoid debris from a 2007 Chinese ASAT test. China's Long March 6A upper stage has exploded three times in the past 4 years, creating thousands of pieces of debris that will remain in orbit for decades.
If you go through Wikipedia's list of space debris producing events, you'll see that zero of them involved SpaceX.[2] That's pretty impressive considering that they're responsible for the majority of mass in low earth orbit. In addition to all of their work preventing collisions, they operate their satellites at lower orbits that decay quickly. If anything does go wrong (such as with Starlink satellite 35956), debris deorbits within weeks.[3]
1. https://starlink.com/updates/space-safety
2. https://en.wikipedia.org/wiki/List_of_space_debris_producing...
Many people in this thread have made claims about limitations of frontier models, but I'm the only one who has shared a conversation with one. Everyone else is either sharing conversations of smaller models making mistakes, or they're making claims about frontier models but not linking to examples of them falling over. If frontier models were so easily fooled, you'd think someone would link to a conversation showing that.
Why look at the rate of improvement of free models when you can look at token pricing? Back in 2023, GPT-3.5 cost around $20 per million tokens. Astra costs half that.
The worry is not that smaller free models will replace people's jobs. The worry is that future models will. We are talking about the capabilities of frontier models because those put a lower bound on the capabilities of future models. Extrapolating from smaller models is a waste of time, as you can interact with the frontier model to figure out its capabilities and limitations.
Also the timestamps on the shared conversations show that you asked Gemini 10 hours after ChatGPT, which means you asked it after your comment claiming you asked both models.
You're in the discussion section of a post about OpenAI releasing hundreds of novel mathematical proofs, and you're claiming that AIs can't apply known techniques to modalities & problems outside their training data? I'm not sure what else would convince you.