pedalpete

We don't have a wash here at our office, we just use a small container with IPA.

We have a UV cure cube, but often I just cure under a $20 amazon desktop UV light.

Look at the first image with the people in the field. That looks to me like a very poorly generated image, not a real image of people, and not even a high-quality AI generated image.

Is this what we can expect of the new apple? I'm not against AI generated imagery, but I expect them to be of equal quality to a real photo.

Or am I wrong, and it isn't AI generated?

The things that make me think it is just poorly made are 1) the angles of the laptop on the black case, 2) the lack of any lighting on the back panel, 3) the size of the hand on the woman in the back, and the squareness of what she is holding (if it is even supposed to be another mac).

I'm wondering if we could create one-off phone numbers similar to one-off credit cards when we have to give a number to a service we don't want to be regularly receiving messages from.

I wonder if the entire thing would be better if the first step was for the LLM to try to understand the final goal of the user, rather than just the simple instruction.

For instance, the LLM may have an understanding that the ball is easier for it to pick-up than the elephant (which it seemed to struggle with, which also made it a strange choice for the demo).

How much of an LLM understanding the physical limitations of the robot would help the task be accomplished in a way that satisfies the goal, without following the exact directions.

I'm in Australia

Dentistry needs an overhall in general. It seems antiquated, and I was hopeful when Wally Dental care raised VC funding that this was a new direction for dentistry, but it seems to be just dentist as a subscription.

I've never really had problems with my teeth and don't go to the dentist often. The last time I went they told me I needed to get my wisdom teeth out. Do they both me? No. Not only that, but in order to get them out, because I have deep roots, they would need to chip away at both my jaw and cheek bones to get them out.

I thought the entire thing was funny when I kept pushing for the "why do these need to come out" and the response was, well, without taking them out, we can't guarantee that you won't need to get them out later and that would be even more difficult then.

But the one way we can guarantee we don't need to take them out later is to take them out now.

I haven't been back to the dentist again.

I think data centers and power are the dark-fiber of this current "bubble".

Inference is 2/3rds of the power used in AI, and I expect the majority of inference to go local in the next 2-3 years, about the time these data centers will come online.

The other 3rd is training, and the major labs are training 6-7 times a year right now. But I suspect we will see a few things. 1) training iterations will slow as the improvements between training runs gets lower and lower. Is it worth spending an extra $500M(ish) for the small improvement in quality? Who is paying for that?

2) Training improvements and improvements in GPUs will require less power. Along with fewer training runs, as we add more and more tokens, the amount of power needed for a training run will likely asymptote at some point.

The point is not that these ideas were wrong, but that they were incomplete, yet were assumed to be complete.

I work in neurotech and specifically with EEG, some would find it surprising that I 100% agree with this take.

When we understood the world of fluid-dynamics, we looked at the body as blood going through a pump delivering materials (oxygen, minerals, etc) to different parts of the body. We though to ourselves "we know how this works!".

Then we discovered electricity, slapped our collective foreheads and said "oh! It's electrical!" now I get it.

We still don't.

Let's learn what we can from what we can and use this knowledge to our benefit, but let's not assume we understand the underlying mechanisms just because we can find correlations.

I'd add that we've seen how the leadership/reputation of different AI leaders, and with 2 decades of history with Larry and Sergey, they seem to come up on top as the most respectable.

Did they ever really abuse the power Google could have wielded? I could be missing something, but for the most part they seem to just get down to building and pushing technology/science forward and avoid drama rather than welcome it.

Let's not ignore the economy where the agent is the buyer.

Video editing and website design being the example here. If the video is an ad, or the website is a sales site, if an agent is a buyer these change considerably. That is an entirely new economy, similar to how we did not understand the knowledge economy until it was fairly well developed in the mid 1900s (someone smarter than me might have better info here).

I have no doubt the economy will change, but we as humans operating in the world will still have needs that need to be met.

In addition, with a shrinking population in all developed nations, we may well still have a shortage of workers in 30 years, not an oversupply.

From what I could see in the blog, pneedleman asked AI about the perfect alarm clock, but didn't ask his daughter.

Did the daughter get involved in the design/build process at all?

It seems like a great opportunity to engage the daughter in some way. Better than buying off Amazon, better than just having AI do it.

Maybe I'm wrong or missed it in the blog, but I didn't see any reference to the daughters involvement at all.

An interesting thought experiment, but I'm not sure I follow the reasoning of these two statements, which is the crux of why open software fits this mold.

1. Reproducible, meaning executable, as well as modifiable, and 2. Reliable, meaning that the results are consistently trustworthy

Why does modifiable fit reproducible. Modifiable fits the open source theme, but not the science theme. Can someone set me straight on this if I am missing something?

Number 2 is something I've been thinking about more recently, and the definition of science from the post being "testable hypothesis and predictable".

Medicine is testable and predictable, but not reliable. To me reliability is the job of engineering. To take what is known in science and make it reliable.

I used to work at CSIRO (Australia's science and technology org) and I remember my boss telling me that researchers always think their work is ready to go from the lab into the real world, but it is our job as engineers to make that translation so that the work outside of the lab. That's what I think about when I think of reliable.

Thoughts?

I was hoping to see something like this, though I didn't know what form it would take.

A few months ago I was at a demo night in SF and 4 people demo'd essentially the same tool for managing PRs. At the time my thought was "are we really going to be dealing with PRs in 2 years.

I don't know if Delta is the right solution, but Zed is at least questioning the current paradigm, which I think is important.

Part of the question may be how do you change the users expectations.

If you are building a product the user already knows, that might not be difficult.

We work in the sleep space, where everyone expects a hypnogram and a sleep score. Even though these things don't actually tell you anything real. They are of little value to most people, but that is the expectation.