[pgodzin]

> Not long ago, for example, while sitting with me in a cafe, my 3-year-old daughter spontaneously realized that she could climb out of her chair in a new way: backward, by sliding through the gap between the back and the seat of the chair. My daughter had never seen anyone else disembark in quite this way; she invented it on her own — and without the benefit of trial and error, or the need for terabytes of labeled data.

I really hate the constant comparisons of AIs to babies. The author's 3 year old daughter has had 3 YEARS of sensory data obtained through moving and trying to fit through things. That is terabytes worth of data! I would expect an AI to be able to generalize once as well.

[zebrafish]

Yes terabytes.... Probably petabytes or more. And all of it is unlabeled data.

If you fed video and sensory data to a deep net for 3 years and somehow were able to come up with an activation function that modeled "survival", I still highly doubt that anything at all would come out that remotely resembles human intelligence. There's no way that i'm aware of to label reality in real time.

[cbames89]

One way to label is by expectation. If the outcome is what you expected, then it gets one label. If it didn't it gets another label. Deepmind has had some success with this approach.

[mehh]

Culture labels things, and we are trained with that.

[sebgggg]

But it is. Pleasure and pain are hardwired.

[0xbear]

Not to mention that kinds don't even know how to _see_ when they are born, and for several months thereafter. That's petabytes of visual stimuli going through that brain so that it learns what's statistically more important and infers semantics around it all.

[Eridrus]

Not to mention millions of years of evolution for initialization.

[dlwdlw]

I think an engineering based mindset of perfection creates brittle models that get stuck in local maximums. Reverse gravity and you can think of current AI as small cars getting stuck in pits. The desire to maximize leaves no room for the human equivalent of refactoring. Giving up the good and going through pain in order to reach new heights.

Instead of an anti fragile model that can continuously be run, current models need to be scrapped at error states. Hyperparameter tuning is just random guessing. Getting good data also doesn't work because of the high dimensionality of it for non-trivial tasks.

The idea that ctrl+z is good should be re-examined. Perfect memory like block chains have isn't the answer either though. Perhaps something similar to the non forgetting yet imperfect human mind.

Most data is garbage and even more eventually becomes garbage. (unless you exist in a finite defined world like Go) Is there any sort of neural net that that find or creates "core" memories with weaker supplementary memories?

The quintessential experiences would only be dislodged with an influx of contradictory data. Initial cores could be initialized via mother-child like training. The training data would be tiered and weighted. There would be an internal system that passed judgement on new ingestion sources. New data would be a necessity. Old data passed in as new would be like a monotonous life digging in cores preventing them from having meaningful change. Almost all data would be labelled as garbage initially unless vouched for somehow. Pure good data would be bad as well because there isn't enough quality differentiation to see what is core and what isn't.

[cupcakestand]

> I really hate the constant comparisons of AIs to babies

Constant comparison? First time I see this comparison to a child.

And btw, it's about that we always equal AI with machine learning, pattern matching, etc.

To be as smart as a 3 year old, we need something entirely different, something way beyond machine learning and all the as AI classified techniques we are aware of today (I think this is what the author meant).