Comments by "sharper68" (@sharper68) on "Sabine Hossenfelder" channel.

Solved until they make the patent secret.
2
But it is not meaningless. There are very specific meanings depending on the application you referring to.
2
@iseriver3982 If you say so, the fact is there are AI applications right now. That they can not chat with you does not change their nature. It is your definition of AI that is limiting your perspective. All forms of intelligence have distinct limitations set by their supporting structure, ability to access information, speed of processing and set of inputs. That current computer AI is scaled back in function compared to a biological version does not mean they are not "intelligence" engines despite limitations. You are basically saying 8 bit games are not games because we have much better versions .
2
@iseriver3982 If you say so, this speaks to your lack of understanding of what AI means and little more. By your spin no AI will ever be more than "algorithms" because no matter how complex or capable that is what they will be built with.
2
@iseriver3982 So? That it is a rudimentary form of AI does undermine the value or the potential of a product that is still in a nascent form. Dismissing actual AI as out of reach because because of limitations of our current hardware ignores how fast our tech has changed or what is in the pipe.
1
@iseriver3982 That means little, talking about the fact we do not have the tech right now to implement a Turing test AI does not mean there are not scaled back AI applications in production now. The "algorithms" you dismiss are part of the same generational development cycle that saw the advancement of operating systems and easily accessible development tools. The current "AI" in development now are the building blocks of what will one day be a science fiction AI. That the current applications are not as sexy or flexible as you seem to want does not change their nature. Mono task forms of any "intelligence" appear rudimentary when compared to what we consider actual intelligence. But their creation and use is still a critical development step along the path to a more recognizable instance of what you personally consider AI.
1
@iseriver3982 Learning algorithms are a component of AI and not the whole show. We have engines that we task to evaluate large swaths of data. They take advantage of processing power and inputs we can not apply ourselves. There are interesting applications like the translation tool written that recently decoded the Voynich manuscripts. This was a document that confounded historians, translators and cryptologists for over 500 years. It did so by recognizing a similarity to an obscure dialect and form of writing never before connected with the papers. Any application that requires a complex deterministic evaluation of large data sets will be ever more handled by these types of mono task AI's written to do what our minds can not.
1
@iseriver3982 Argumentum ad populum, very convincing. Hate the term if you want to but it has a distinct meaning that applies to the applications I am referring to that you choose to dismiss. You started the post stating you did not understand the meaning of the term and hammered that home.
1
@iseriver3982 I am saying you apparently arbitrarily dismiss our AI development because they are constructed with "algorithms". Saying I am wrong because of list pulled up in a google search of those who agree with you is literally arguing from popularity. No matter how you cut it the term AI is not meaningless, it appears you just do not know the meaning.
1
@iseriver3982 Any AI that exists or ever will needs to utilize a series of algorithms as its core. The difference in our perspective is our expectations of what the word means. There is a market element that sells inventory management systems that factor in historical data evaluated through an algorithm to suggest restock levels as AI. I understand the argument these "algorithms" are not AI in the star trek Data sort of way but they utilize decision tree mechanics inherent in more complex logic engines. It is arguable they are all AI of varying scale of complexity. The finished product you will admit is an AI will just be a more complex implantation of the concepts of autonomous learning and data logic theory we currently work with on hardware that meets the demands you make of it.
1
It will happen the same way everyone converted to electrical lighting and other appliances as they too drove the need for more production. Demand drives the update and that they already have unattached capacity means that the money will be found to fill that gap. I am unconcerned about their ability to catch up to a capacity problem as it has always worked out in the past, the road blocks now are largely bureaucratic. This is the kind problem solved with will. All that said Hybrids are a good option for many, the electric car is a good short commute platform. You will have plenty of electric to charge both.
1
That even AI's run faster on quantum machines means quantum machines are not going away.
1