[MikeB]

This is trending #11 on YouTube.

Speech has a solid three second delay, and responses are what you expect from the statistical analysis and middle-average contextual analysis approach.

It would be a whole new layer of interesting if responses included the own robots' survival instinct, like throwing in a few "what about me?" and "have you seen my charger?" responses, but maybe that's a bad look or conclusions could be drawn about the danger of AI in a physical sense.

[8pla.net]

A Few Constructive Criticisms:

In my opinion, it seems...

• Everything (robot, table, items) has a fixed location.
• The stimulus is scripted for keyword recognition.
• The responses are simply a generative AI application.
  

These criticisms are not complaints about this video.
They are for chat purposes about how this may work.

Are my opinions, correct or incorrect?  What do you think?

[ivan.moony]

Are my opinions, correct or incorrect?  What do you think?

I believe you're correct about fixed locations, but not in extent to pre-scripted actions, yet the bot may be super-trained to shown specific inputs/outputs. It is yet to be seen how the bot behaves in stressed, unknown conditions.

But overall, that kind of bot body seems like a big step forward, reserved only for Boston Dynamics by now. Training on what to do with the body may be mere formality for ANNs.

Is anyone aware of what software tech Boston Dynamics used for, say, Atlas?

[MikeB]

The items are not in a pre-programmed/fixed location. Advances have been made in machine vision, so basically anywhere nearby in view is good enough to identify the item. Eg. the physical appearance. Whether the item is relevant to contextual conversation is another matter.

But the engineer in me shudders at the thought of comparing millions of images just to identify the probability of an item in view....

Advanced, but can't say any advancements have been made