Kitty hawk. 747s carrying millions of pounds. Moonrockets. But because we haven't produced the common barn swallow with lungs and a heart and blood pumping through its veins, the entire aviation industry is a scam and we can conveniently ignore it?

"It's just predicting the next token". Yes this has been well known for a while now. Is there a serious question?

The internet has been proven to convey negative information farther than positive. We see this with people predicting the demise of the stock market. So why not some opportunistic folks saying that well, it just predicts the next token. To use the word "scam" may be a projection? Like this hand waving will actually take the problem away.

A 1 dimensional system solves a lot of problems that the proposed 3 dimensional system has: Instead of sending the exact location of each node, it broadcasts a much simpler on-demand beacon with distance determined by multi-path resistant RSSI. 2.4ghz would be used. beacon and beacon/inquisitor modes. 8 node id types (car, bike, pedestrian, hazard, emergency vehicle emergency/non(even more important because people forget to set emergency mode but nearby car AI sees it moving fast and concludes it is an emergency), etc.). 4 bit random locally unique id.

1. tamper resistance. this is a backup to conventional cameras/radar. As a result, it is not mission critical, is optional and there is little incentive to spoof it. It helps in camera blind spots areas that spoofing is least valuable, rapidly disprove (just wait a few second to see it is false and then ignore) and the car would be going slower anyways (second car ahead, oblique side, etc.). Also more jam resistant since adaptive and no GPS required.
2. Accuracy. in simulation, even with simple RSSI distance, there is an interesting effect where camera view verifies most beacons and therefore obstructed beacons get sharper in location since there are only so many blind spots at at each distance.
2. privacy. The id is only locally unique and beyond each encounter does not uniquely identify the node.
3. cost. No GPS required. the node uses common 2.4ghz transceiver tech. modified wi-fi like ESP-now? these frequencies are underutilized in the streets according to rf maps. Urban range is less than a mile.
4. battery life. in beacon mode, the battery life would be years. node is very small, cheap and long battery life, encouraging more people to use it. in vehicles, it might added merely by applying a software update, no new hardware required.
5. Adaptive. inquisitor mode can temporarily increase beacon response rate or go quiet in areas of good visibility. beacons automatically switch channels as needed for congestion. beacons only transmit when prompted by any inquisitor packet and then only to max rate and duration.

Manufacturer liability for this product should be less because it's not designed to be the primary locator system. It is a backup system.
This is like wearing reflective clothing. you still have to be looking for pedestrians, etc. but this makes it easier to see them.
Combine with normal internet sharing of road hazards. Location of weird beacons (someone threw a pedestrian beacon onto the freeway), etc.
Other applications: turn on AC as you walk towards car. garage door. Forest service station location. etc.
extended protocol supports short messages/URLs to be broadcast with beacon or given by request.
Suggested product name: nearSense

Smart phones, because they use shared RAM for video means that most people are more likely to be able to run a LLM on their phone than on their desktop/laptop. And local LLMs are essential to the myriad of applications that can be unlocked because of privacy like medical, etc. It also allows more nested prompts (mild AGI).

Truly democratizing AI (ownership) is the best alignment strategy I know of. "When everyone is botting, no one is botting."

see post below agreeing with yours and going into more detail, "removable batteries?"

- Improves the car to lithium mined ratio by greatly increasing battery utilization
- improve MPGe to over 150 (100-130 is norm for sedans) for full size vehicles for typical driving cycles (stop and go are more sensitive to pack weight vs aero). This is a big deal, improving mpge by 20% accelerates a virtuous cycle like improving specific impulse does for rockets. It can greatly reduce vehicle costs, charge times, energy use for miles driven, etc. Biggest effect might be with trucks. They have a wider range envelope because of payload use. A large truck with modular packs would be able to both reduce costs for most driving yet retain tremendous range for long/heavy trips.
- reduces an already lower maintenance cost for EVs by reducing tire wear (number one maint cost for an EV). For 95% trips, a comparable vehicle would weigh less than the ICE equivalent. trimming 600-800lbs from sedans and >1000lbs from trucks. again, for most trips.
- longer battery life (batteries fail as outliers, modular packs reduce the effect of outliers)
- more upcycling vs just recycling. significant after market for a standardized pack. Used packs are still good for less energy intensive use in home storage, RVs, motorbikes, side by sides, tools.
- reduce charge time for long trips (swap and go is faster than fueling an ICE). More time allowed to condition packs in charging station instead of in car.
- Smaller/cheaper/more plentiful charging "stations". Like the propane exchange cages at the local quickie mart. Cassette system charges/conditions/tests packs.
- Improved charging station surge capacity. All packs for a rush period don't have to be charged in that rush period.
- increases competition for the most expensive component in a car.

This will heat up in the next 5 years:
- at current organic rate of improvement in the battery industry, we should see a 40lb, 15kwhr (50 mile) pack. industry standard connector for power/data/glycol cooling (or do transfer cooling through housing). Shaped like a blade server.
- Typical car might have have 8 bays (4-6 under passenger compartment, 2 or 4 under trunk and frunk). Sold with 2 packs standard (more than enough per study in OP above). Extra packs can be bought used, rented, borrowed from neighbor (coop), bought with home energy system, etc.
- one of the reasons for larger packs even though they are not used is for the peak energy capacity of the system. a larger pack can deliver a higher peak energy (better 0-60 acceleration), Smaller packs can do this if we use a mix of lifepo4 for the base and regular lithium for the road trips (standard packs in 2 chemistries available).

Exceptions are a commonly used tool of marketing "you need this extra capacity just in case". But stats don't lie. Most driving would really benefit from less range.

btw, been driving EVs and hybrids for over 2 decades. It is full every morning. don't have to wonder if someone forgot to get gas. Always ready for a quick out of town trip if needed vs our ice we often had to stop as the gas station first. Our driving is pretty typical and for us, we spend less time "filling up" than we did with our ICE. Figure 30 trips to the gas station each year at 10 minutes each vs 3 seconds each for my EV plus a few public charger uses for our out of town trips. And when we do stop, it is a bit more interesting of a place than the corner gas station.

Yes, I am surprised more people haven't said this.

Truly democratizing the technology is the most practical solution to the "Alignment Problem" I have seen.

They realize they have no moat (read the leaked paper). 1. encourage the government to "regulate" the industry so they can keep their monopoly. 2. release an "open source" model that inexplicitly neuters the open source community

When a monopoly claims to be pro-regulation they know full well this will hurt their competitors more than them. See "we have no moat" about how Google and "Open"AI have realized open source is coming for them. Regulations will save them from us.

Usually we have a different set of regulations for development vs production. Ethical lab trials, Product liability, etc. Normally our legal system is a lot more permissive of what you can develop vs what you can sell to the public. Part of the problem is that as with dangerous pathogens, there is a very thin line between development and "production" with AI. So in order to regulate production, it looks like regulation of development. Regulating science and even thought. This is dystopian. All a mute point if we just promote a system were it is very democratized. "If everyone is botting, no one is". This is the best form of alignment. the most reliable way to address the job loss issue, etc. And we have run the experiment before (the '75 microcomputer revolution vs the mainframe (cloud)). So the best regulations for AI? keep the ones we have and discourage monopolies like "Open"AI trying to use the old and tired, "regulate open source so we our lobbyists can exploit the system" trick.

Ah, reread it again. You are right. He dissented.

""I know the thought process is, 'Just trust us, just trust the police department.' But the reality is people have been violating civil liberties since my ancestors were brought here from an entirely, completely different continent," Walton, the board president and District 10 representative, said." Self-refuting logic?

Ostensibly because grandpa keeps calling their tech support. This will further drive power uses away from their operating systems. PC's represent one of few remaining islands not centrally controlled by the cloud. It wasn't the mainframe that made the Internet what it is today. It was the mainframe that produced all these new types of software and many jobs. The cloud is the new mainframe. It will further consolidate the market. be used to cut costs instead of add features. Cutting jobs instead of adding them. Throw AI in the mix and we have a powerful tool being centrally controlled. If we want to see new applications with AI never imagined by the corporations, it needs to be like the microcomputer revolution of '75.

that could said for a lot of beliefs...

We have dealt with clearly defined ecological disasters in the past and they were and continue to be best solved through the most libertarian systems of government.

lead gasoline. background lead levels very high. birth defects. etc. Lots of governments read the report and wanted to do things. Governments more socialist. they certainly had the ability to just force people. but holding on to power was more important. No, it took a more free country to get the ball rolling for the rest of the world.

cfcs. Hole in the ozone layer, increased uv, etc. Again, other governments have smart people and they had good intentions. Lots of young people who wanted to be our next savior. but they lacked liberty. Once again, it took the most free country in the world to lead the charge on this.

Do we think we can now do better with clearly defined problems by using the styles of government that have failed to do as much about them in the past?

And the article (read the original on vice) is a joke. They predicted something far in the future that still hasn't happened. but then they say it is on track because some trends they characterized in the real world are still there. The main problem is that their claims are not falsable, and therefore, not ultimately scientific. It requires belief.

https://www.youtube.com/watch?...

As Pinker, etc. have shown, we do have surprisingly good trends. We do know systems that are not just predicted to work in some imagined utopia but have worked in the past. Yeah, wish they were faster. But many of the new proposals would make it worse.