Sam Altman argues top AI scientists delayed progress because their personal identities were tied to scaling skepticism

@ihtesham2005 "The smarter you are, the more confidently you defend the wrong position."

Almost - this is describing mid-curve smartness. Truly smart people can admit they know very little and remain curious.

Also - people lacking smartness exhibit the same behavior (Dunning Kruger effect)

@ihtesham2005 "When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong" — Arthur C. Clarke

@NaveenGRao I remember folks scoffing and outright laughing at your projections of DL training market..

@ihtesham2005 I was thinking about this : This idea must die https://www.amazon.com/This-Idea-Must-Die-Scientific/dp/0062374346/ref=zg-te-pba_d_sccl_1_3/137-3316220-7226321?pd_rd_w=zE6d9&content-id=amzn1.sym.081392b0-c07f-4fc2-8965-84d15d431f0d&pf_rd_p=081392b0-c07f-4fc2-8965-84d15d431f0d&pf_rd_r=PRHHE3F7Y7XKEAJ2A91G&pd_rd_wg=vi06r&pd_rd_r=87e43456-df6a-4754-8dfc-cb460fc6c884&pd_rd_i=0062374346&psc=1

@ihtesham2005 The data shows that scaling does not lead to general artificial intelligence

@ihtesham2005 I see it online all the time. “Hi, I’m Dr. Knows-Everything with a PhD in Better-than-you. I’m going to cite outdated sources for my arguments, straw man yours, and then play the expert card when I’m called out.”

Intellectual humility is separate from intelligence.

@ihtesham2005 this is the same for all those getting butt hurt by ai replace their jobs, as for many their job is their identity

@ihtesham2005 Altman misunderstands the beast that his company created. Those scientists are right. Scaling hasn't produced AGI and never will. It's produced a massively overfit and poorly generalized model. It just happens to be trained on everything.

@ihtesham2005 George Carlin did a skit on this same topic:

https://youtu.be/dDJFH358dRg

@ihtesham2005 This happened in Astronomy in early 20th century. When young Chandra proposed his finding, Eddington resisted that with all the power he could muster. https://en.wikipedia.org/wiki/Subrahmanyan_Chandrasekhar

Chandra X-Ray Lab is named after him.

@hanlintang Yeah…exponentials are hard I guess

@ihtesham2005 If done right, things scale. If a poor model is followed , obviously there will be issues. Sometimes the best new ideas come from outside a field or an adjacent field, not blinded by preconceived notions.

So are we going to pretend Hassabis and Illya are not the ones literally writing the book on AI?

While Sam was spending 7 years on Loopt, and Murati was working on the failed tesla falcon doors, those guys were inventing neural nets and inventing alpha go.

In my opinion, the field is held back bc most investors are charmed by the non experts in fund raising, while the smartest experts generally don’t love the roadshow

@ihtesham2005 Tech need engineer to lead... he's not one of them.. He know how to sell dreams to investor..Scaling for him it's not issue , because it's his selling point , instead of hire more scientist you build more data center.. the later get money easier from investor

@ihtesham2005 Is this a diss at Yann Lecun?

@ihtesham2005 The irony is he is also describing his own stance on AI.

@ihtesham2005 Interesting that most of the 'non-experts' are the ones who can figure out how to reset their iphones. I'm not sure why we would have so much faith in rhetorical lay thought, particularly from those with bad taste, other than to fuel intellectual grandiosities.

@ihtesham2005 10 years ago Machine Learning was going to solve Climate Change. It is difficult to argue for that now if your approach is the scale-is-all-you-need ideology.