I'm increasingly starting to think this is the end-state of AI. The internet becomes infected and fundamentally untrustworthy.
At the moment, the current frontier models require significant infrastructure to run, so I'd like to think we could locate and contain swarms of nefarious frontier models. However, if these models can understand how to federate themselves into more distributed networks then that containment becomes questionable.
KGB's agents are human, OpenAI's agents are not. It's an important distinction because humans are responsible for their behaviour, while AI agents are not.
You cannot try an AI agent in a court of law, despite the anthropomorphising work the word "agent" is doing.
Exactly. It’s still just software, which someone programmed and deployed to do specifically dangerous/malicious things. I feel like we already have legislation and case law surrounding this.
AI epitomises chasing a narrow definition of progress that benefits the few (increasing profits via automation) over a holistic definition of progress that benefits the many (reducing poverty, improving health, providing meaning). It's no wonder people hate it.
The Claude Code extension in VSCode is similar enough. I used to use Cursor, but now I use VSCode+CC to take advantage of the better pricing CC offers for always-on frontier. I found Cursor's affordable "Auto" option unreliable, often wasting my time with dumb models.
For the kind of ML work I do Cursor's "Auto" was too unreliable. It would frequently try to solve relatively difficult tasks using Composer, when it should have been routing to a true frontier model. Annoyingly, doesn't tell you what model it's decided your task deserves, so I often wasted time working through a problem with it, only to realise I was on a dumb model. Then I revert all the work, switch to an expensive frontier model and pay expensive API prices to actually solve the task.
All that is to say, model selection is the main control we have over quality. Giving it up in the name of cost saving will bite you in the long run, especially when Claude Code still has such good plan pricing.
I also had a problem with Cursor auto switching on fast mode which I would never use otherwise.
Cursor really needs a stronger model than Composer 2 which is the root of the problem. Hopefully they will put out something better now that they have better datacenter access with SpaceX.
I do this with my team. We spend more than half of our weekly 30 minute 1-on-1s talking about anything but work. That isn't written down anywhere, it's just a natural consequence of us being interested in each others' lives, and prioritizing that over "getting back" 20 minutes to do more work.
We also have a team-wide monthly "happy hour" where we bring one discussion point each, usually an interesting article. They're a lot of fun, and I appreciate my colleagues in a much more rich way than I would have otherwise.
It's so obviously important that we maintain semblance of community through live conversation in remote workplaces. I spend more time "with" my remote colleagues than I do with anyone else in my life, including my wife. The human brain does not separate cleanly into "colleagues" and "friends".
What if they don’t want to spend their time making small talk with their boss or manufactured chit chat at a happy hour?
I get that a lot of people need this. I don’t need my work to provide a social life, I’d rather get my shit done and leave. When I socialize in a work capacity I’m doing it because it helps me do my job better, not because I want to
If they don't feel like chatting they don't need to. I guess I have a chatty team. We don't always fill the 30 minutes, but we often do. It doesn't feel performative and I wouldn't waste our time if it did.
There's no consequence to skipping the happy hour but everyone shows when they can and is engaged and contributes to the conversation. We often talk about culture, politics, economics other topics most folks have a general interest in.
> I don’t need my work to provide a social life
Fair, not everyone sees work as a community, but many do. Remote culture is hard to build but it needs to allow for folks who still need human connection in their workplace.
The risk I often see is when the company also emphasizes this 'family' ideal. I think that is unnatural and forced; most of my associates hated this. It ruins the work-life balance.
I found it more important to emphasize trust, and allow them to handle these conversations/attendance
If they couldn't, that's fine. Outside factors can disrupt this, ... So I wouldn't complain if there was a no show once in a while.
We had a monthly tea(m)time to share tea and talk about anything, hobby topic or something technical. It was fun to see what people do with 3d printers, especially those that had no time/space for this.
This isn't a company policy, it's just something I make space for in my team. There's no pressure to connect beyond the required work items.
The happy hours are only an hour a month during work hours, where we talk about stuff everyone on the team is interested in, like economics, culture and politics. Most people seem engaged and bring interesting stuff to talk about.
We also do a more formal "data blitz" in our ML team labs, where folks bring vaguely data-related topic to talk about for 10 minutes, but that requires a little prep. Those are always fun. We've had people rating diaper brands and showing their marathon training schedules.
I've had 1:1s like this and I'll tell you it was entirely performative. Yes my manager and I chatted about non work stuff for half or even all of the meeting but it was all fake, just talk to get through the meeting.
It's not like we try to fill the 30 mins, but I think we all crave a little connection with our colleagues, so make use of the time to catch up. I've worked with these same team members for over 3 years, so we are all a little invested in each others' lives by now. Obviously I'm not going to force a new team member to divulge everything about their lives, or go through my weekend in detail unless they're actually interested in chatting. It's pretty clear when someone just wants to get on with their work, rather than chat.
There are various levels of self deception that almost everyone subscribes to.
A pretty high level one is that our jobs are meaningfully making a positive difference in the world, when in fact, most white collar jobs are just producing bullshit to grease the corporate wheels of modern society. Most people don't like to admit that though, so we tell ourselves little lies and go along with the corporate narrative. That's what you experienced.
But it goes deeper the more truthful you try to be. Down near the bottom of this pile of self deception is that humans are making the world a better place, when in fact we're ruining the world, causing environmental damage at an unprecedented rate in geological history, all the while exhausting the readily-accessible non-renewable resources, like hydrocarbons and minerals, that'll make the chance of a better future civilization on Earth highly unlikely.
This is a bit of weird article. On one hand, I understand what they're getting at: AI is a transformative technology, but the people whose lives will be most transformed aren't included in the conversation. On the other hand... of course that's how it is while AI is in the hands of literal profit seeking corporations. That won't change until the labs are nationalised under a government that cares about its citizens' wellbeing. One might counter that a good corporation will listen to its customers, but that has never been the case for powerful technologies with real costs for users to not adopt them.
The article is talking more about people like translators being replaced by AI translation. I don't think any of the labs have a department of making it worse so it can't do people's jobs.
The normal way of dealing with tech doing peoples jobs is to help them get different jobs. I've got a translator friend who did a government paid course to train as a tour guide - that sort of thing.
Dario and Demis have called for nationalisation at some point. They know if AI reaches what they believe its potential to be, it needs to be democratically governed. It will upend the markets, but AI already threatens to do that. It feels like wishful thinking given how entrenched we are in neoliberalism, but it makes sense.
In the mean time there are various avenues of regulation and redistribution to lessen the effects, including retraining programs, though that job creation will keep pace with job losses is a big unknown.
At the moment, the current frontier models require significant infrastructure to run, so I'd like to think we could locate and contain swarms of nefarious frontier models. However, if these models can understand how to federate themselves into more distributed networks then that containment becomes questionable.
reply