Rebuff5007

> Moving fast and breaking things is fine, as long as you fix the stuff you break...

What? No its not. Breaking things can cause harm that is not always "fixable", particularly if its not your thing to break.

"All models are wrong, but some are useful"

What your describing is already how a lot of science, technology, and engineering works!

Between this comment, and the comment above, I dont know what feels like fair criticism here.

Having a single perfect product strategy with non-overlapping product categories and understandable names is hard for any organization, particularly in a rapidly evolving space.

Its obviously an issue to have multiple mature products be chaotically names.

At this moment antigravity and gemini cli and are hardly mature. Isn't now the perfect time to consolidate?

I'd argue that the engineers of 20 years ago were better than the engineers of today because they were significantly more resource constrained and for example, would never use a 300mb javascript library for a profile page.

Not sure why this is hitting the home page right now but people may also be interested in Mujoco Playground [1] which is the latest RL environment wrapper of mujoco, implementing both classic deepmind-control benchmarks, and some very new interesting ones!

[1] https://playground.mujoco.org/

Snowdens document leaks happened in 2013 (implying the surveillance state was set up well before then). So this is more a leisurely stroll than a sprint.

I see your point, and think its valid, but here is a counter:

Content is graded on both instant appeal (e.g. rotten tomatoes "popcornmeter") and artistic appeal (e.g. rotten tomatoes "tomatometer").

I firmly believe that AI generated content cannot have any artistic appeal, because I believe art is fundamentally an invocation of human expression. This might be fine in some contexts, but in general I'd prefer consuming content from groups that I trust to strike a good balance between these types of appeal (e.g. A24 movies).

Very interesting discrepancy in the attached example:

- "removing the kettlebell" led to removing the visual representation of the kettlebell as well the deformation it makes on the pillow

- "removing the hands" removed the childs hands from the tops, but did not then lead to the tops falling over!

Others like the colliding cars are in some weird gray area between the two.

One should note as these tools proliferate, there is a lot of artistic expression that we are giving up to these imprecise natural language parsing engines.

> I bet OpenAI genuinely believes

What does this even mean? Who is being "genuine"? This is far to naive a take for a company thats burning through hundreds of millions of dollars, and constantly striving to set the tone of AI and their own supremacy.

Of course it is... we are in an era where a well-timed blog post showing "SOTA results" on a benchmark can net millions in funding

I share this dilemma too. Just a thought -- I feel less okay with AI processing "data made for humans" (i.e. the images themselves, audio recordings of speech) and more ok with it processing "data made for software" (exif data, shazam logs).

The UK government is trying to regulate a service used by UK citizens / residents. What part of this seems unreasonable?

I dont think this is right... most people I know care more about not doing the "wrong" thing than feeling entitled for doing the "right" thing.

> it's very hard to see how anyone could look at what just happened

I think what you are missing is their annual comp with two commas in it.

I worked in some fully autonomous car projects back in ~2010. I would say every single company and the industry at large felt HUGE pressure to not have any incidents, as a single bad incident from one company can wreck the entire initiative.

Do you also audit every part of every car you buy or medicine you take? Or do you rely on large well-established institutions to do that for you?

"Dont trust google" imo is the wrong response here. We are at the mercy of our institutions, and if they are failing us we need mechanisms to keep them in check.

I've also been quite skeptical, and I became even more skeptical after hearing a tech talk from a startup in this space [1].

I think the best way to think about it is that its an engineering hack to deal with a shortcoming of LLMs: for complex queries LLMs are unable to directly compute a SOLUTION given a PROMPT, but are instead able to break down the prompt to intermediate solutions and eventually solve the original prompt. These "orchestrator" / "swarm" agents add some formalism to this and allow you to distribute compute, and then also use specialized models for some of the sub problems.

[1] https://www.deepflow.com/

OF COURSE!

The fact that tech leaders espouse the brilliance of LLMs and don't use this specific test method is infuriating to me. It is deeply unfortunate that there is little transparency or standardization of the datasets available for training/fine tuning.

Having this be advertised will make more interesting and informative benchmarks. OEM models that are always "breaking" the benchmarks are doing so with improved datasets as well as improved methods. Without holding the datasets fixed, progress on benchmarks are very suspect IMO.

I've heard a lot of such anecdotes. I'm not saying its ill-intentioned, but the skeptic in me is cautious that this is the type of reasoning which propels the anti-vax movement.

I wish / hope the medical community will address stories like this before people lose trust in them entirely. How frequent are mis-diagnosis like this? How often is "user research" helping or hurting the process of getting good health outcomes? Are there medical boards that are sending PSAs to help doctors improve common mis-diagnosis? Whats the role of LLMs in all of this?

Few things that go overlooked from someone that works in industrial robotics:

- note that google already has investment in industrial robotics with intrinsic ai (that is the confluence of OSRF, bot and dolly, and a few other robotics/ai companies)

- this partnership specifically seems to be focused on humanoid robotics, which is not taken seriously by industrial manufacturing folks

I generally don't place "conspiratorial" motivations to explain the decisions of tech companies, but I can't help but think this is a move from apple to keep their ludicrously powerful M-chips busy with mundane UI because the average user will never need to otherwise upgrade past an M1.

Well, they picked it back up with Jax...

Hot take: they have always been firing on all cylinders. The marketing it just a bit different now. Everything you mention is the result of significant long-term investments.

Hot take: i'm not sure this is actually bad for actual diversity and inclusion. From my experience (not at msft), companies continue to have many internal goals related to equal pay, gender balance, etc.

Whats changing is how this is communicated externally, and I can see why this would have to change based on the political climate.

Both of those sound like expertise in building a website, and not like expertise in business strategy.

To be clear, I would personally have a similar view to the author here. I'm just surprised that they think their opinion on the strategy side matters so much to their client!

The expertise offered here is "how to build a website". If the client is insisting that the dev use a specific javascript library, that would be odd.

The client here is just requesting specific content on their website, similar to someone requesting a granite countertop in their kitchen; that seems fine, even if its not particularly classy or aesthetically pleasing to the contractor.

This whole post is coming of a bit naive to me... I highly doubt this client is just an inspirational design meeting away from changing their offering and make a massive investment in customer support. I also don't get why a web-development consultant would feel so responsible for a pretty typical business decision.

Thats true, but I think the blame is more on "American society" and not the kids working through the system.

50 years ago, college was cheaper. From what I understand getting jobs if you had a college degree was much easier. Social media didn't exist and people weren't connected to a universe of commentary 24/7. Kids are dealing with all this stuff, and if requesting a "disability accommodation" is helping them through it, that seems fine?

But why not? AI also has very powerful open models (that can actually be fine-tuned for personal use) that can compete against the flagship proprietary models.

As an average consumer, I actually feel like i'm less locked into gemini/chatgpt/claude than I am to Apple or Google for other tech (i.e. photos).

I think the good news is that open-source models are a genuine counterweight to these closed-source models. The moment ads become egregious, I expect to see and use services for an affordable "private GPT on demand, fine-tuned as you want it"

So instead of a single everything-llm, i will have a few cheaper subscriptions to a coding llm, a life planning llm (recipes, and some travel advice?). Probably it.