This strikes me as a skunworks project to investigate a technology that could be used for autonomous vehicles someday, as well as score some points with Sundar and the Alphabet board who've decreed the company is all-in on Gemini.
Production Waymos use a mix of machine-learning and computer vision (particularly on the perception side) and conventional algorithmic planning. They're not E2E machine-learning at all, they use it as a tool when appropriate. I know because I have a number of friends that have gone to work for Waymo, and some that did compiler/build infrastructure for the cars, and I've browsed through their internal Alphabet job postings as well.
I have a similar school drop-off, and can confirm that the cars are typically going around 17-20mph around the school when they're moving. Also that yes, human drivers usually do stay much closer to the centerline.
However, Waymo was recently cleared to operate in my city, and I actually saw one in the drop-off line about a week ago. I pulled out right in front of it after dropping my kid off. And it was following the line of cars near the centerline of the road. Honestly its behavior was basically indistinguishable from a human other than being slightly more polite and letting me pull out after I put my blinker on.
Interesting that they have "IMPORTANT: Assist with defensive security tasks only." twice, once as the very first instruction after telling Claude what it is, and once toward the end.
Somehow I think that the weak link in our government security is at the top - the President, his cabinet, and various heads of agencies. Because nobody questions what they're allowed to do, and so they're exempt from various common-sense security protocols. We already saw some pretty egregious security breaches from Pete Hegseth.
I have never worked in a company where an obviously incorrect CEO-demanded security exemption (like this one) would have been allowed to pass. Professionalism, boards (with a mandatory employee member/representative, after some size) and ethics exist.
30 years in about 8 software companies, Northern Europe. Often startups. Between 4 to 600 people. When they grow large the work often turns boring, so it's time to find something smaller again.
I’m in the US, SE since 1998, startups to multinationals. What the GP said holds true for me too. There are serious professionals in the world - I don’t know why some people want to drag every one else down to the level of the current US administration- they are exceptionally inept.
CTO at a successfull cybersecurity startup I worked at long ago was exempt from critical security updates. She refused to restart her computer out of fear for her Excel state.
I used to work devops for a startup. The _only_ person who was exempted from 2-factor auth was the CEO. It's the perfect storm: a tech illiterate person with access to everything and the authority to exclude himself from anything he finds inconvenient.
Why would you? He’s literally the only person ostensibly in charge of the direction of the company. Destroying the company through a security exemption or a bad business deal - both are the leader making a poor decision due directly to his seat of power.
Give sound advice of course, but ultimately it’s the exec’s decision make.
There are many reasons to deny a CEO ... in a good company structure such denials are circled back around to the board for review.
Case in point: Allowing a CEO with no flight training to "have the keys" to the company <rare, expensive, uniquely outfitted, airframe> because they want to take it for a spin.
Sheparding Royalty in Monarchies has been a neccessary, delicate, loaded, and life threatening role for centuries.
Being a C-suite Groom of the Stool isn't a happy job, but somebody has to do it.
No, it isn't - it's an asset owned by the company and shareholders - a CEO is an appointed or elected officer.
> To be clear, I’m referring much more to CEO/owners
Owners are what you are talking about. CEO / Owners are Owners and can act like owners.
That said, even owners need to be herded like cats when they are making bad decisions that impact tens of thousands of people on the basis of hubris and feels.
Somebody has to toss them shiny keys until the moment passes and they can make rational choices again.
Whoever needs to to do their job. And you put in security controls (e.g. part A and part B). Also compensate your people well and don’t publicize who they are.
Semiconductor does this all the time…engineers on team A know only about their process critical gate materials step. Engineers on team B know about their lithography step. They are trained not to disclose and people respect that.
Been there. The CEO of an internet security company was the one who clicked on the wrong email attachment and turned a virus loose.
I mean, I don't know if he had a security exemption, or if anyone who clicked on it would have infected us. But he was the weak link, at least in that instance.
whether he is personally and directly responsible for this specific incident, his leadership absolutely sets the tone for the rest of the federal government.
Humans generally find "food safety expert sickens guests with tuna salad he left out overnight on warm countertop" to be a far more damning charge than "fire safety expert sickens ... warm countertop".
Dig up a live mic catching Hillary calling the IOC a bunch of self-serving scum just as Obama was begging them to award the 2016 Olympics to Chicago, and we might call it comparable.
Also bears all the hallmarks of an ordinary post (by someone fairly educated) on the Internet. This would make sense, because LLMs were trained on lots of ordinary posts on the Internet, plus a fair number of textbooks and scientific papers.
The — character is the biggest cause of suspicion. It's difficult to type manually so most people - myself included - substitute the easily typed hyphen.
I know real people do sometimes use it, but it's a smell.
I think some software will automatically substitute "smart quotes" for regular quotes and an em-dash for a double hyphen -- I know MS Word used to do this. Curious if any browsers do. This comment was typed in Brave, which doesn't appear to, but I didn't check if Chrome or IE or Opera does.
The comment was not wrong though so I am not sure I understand if flagging it for the sole "it was most likely written by the use of AI" reason is completely valid.
That's sorta what MetaBrainz did - they offer their whole DB as a single tarball dump, much like what Wikipedia does. I downloaded it in the order of an hour; if I need a MusicBrainz lookup, I just do a local query.
For this strategy to work, people need to actually use the DB dumps instead of just defaulting to scraping. Unfortunately scraping is trivially easy, particularly now that AI code assistants can write a working scraper in ~5-10 minutes.
Sure. Development at Google is glacially slow because nobody does any work, and so they're only publishing releases bi-annually because there aren't enough substantive changes to make quarterly releases seem important. This will also allow the teams to move to biannual OKRs instead of quarterly, which lets ICs and line managers do half as much work while giving executives justification for why they need twice as much headcount.
When it comes to large bureaucracies, always assume laziness over malice or strategic competence.
Commuting and residential patterns changed too. A lot of Googlers purchased houses in the Tri-valley during COVID instead of living in apartments in Mountain View or Sunnyvale or SF. Now they have a Dumbarton or 237 commute instead of something Caltrain-accessible. Tech companies also started laying off in 2022, and stopped hiring in the Bay Area; I'd bet that total employment along the Caltrain corridor is significantly lower than in 2019.
The Bay Area also needs way better last-mile transportation. I looked into taking Caltrain to work; it'd take 22 minutes to Caltrain the ~15 miles to the nearest Caltrain station, and then another 22 minutes to shuttle the 2.5 miles to work.
Health and medicine is very far from deterministic. Your drug interaction checker is deterministic because the non-determinism is handled at a higher level (the doctor / patient interaction) in the health care system. Individual patients often respond wildly differently to the same medicine even in the absence of drug interactions.
Similarly the non-determinism in ChatGPT should be handled at a higher level. It can suggest possibilities for diagnosis and treatment, but you should still evaluate those possibilities with the help of a trained physician.
Production Waymos use a mix of machine-learning and computer vision (particularly on the perception side) and conventional algorithmic planning. They're not E2E machine-learning at all, they use it as a tool when appropriate. I know because I have a number of friends that have gone to work for Waymo, and some that did compiler/build infrastructure for the cars, and I've browsed through their internal Alphabet job postings as well.
reply