Artificial Intelligence 2020 - is there progress? - page 72

New comment
 
Rorschach:
russian gpt3
Thank you very much for the link to this article.
 
No need to follow in the footsteps of the Americans. Create your own unique AI technology, because otherwise, what's the point? No self-respecting racer aims to finish second, and if he is overtaken on the first lap, he will do everything to catch up and overtake on the next. The tech race to AI is in full swing, and you're comfortably nestled behind GPT-3, repeating what's already been done. Why? Look for ways to do better. I don't understand why you haven't identified them. If you can make an analogue of it now, you could have done it a year ago, what prevented you? Waiting for the Americans to be the first to come out with their version so they could copy it? What is wrong with you guys? :(((
 
You will not win the AI race, because you are not capable of setting your own standards, but only using American ones. Until they step forward, you will wait and stall in place, looking around dumbly, and then goose-stepping to follow them. This psychology is foreign to me, so, good luck and goodbye).
 
Реter Konow:
There is no need to follow in the footsteps of the Americans.
Reg Konow:
AI won't win the race

I hope you mean Sber, not Russia. Russia is doing quite well in military AI, autopilots and facial recognition, along with the US and China. Google is way ahead in speech recognition. It will be - the security services do not really need it, and there is not much money invested there.

 
Edgar Akhmadeev:

I hope you are talking about Sber, not Russia. Russia is doing quite well in military AI, autopilots and facial recognition, along with the US and China. Google is way ahead in speech recognition. It will be - security agencies don't really need it, and there's not much money invested there.

I'm about psychology and mentality. No matter how crude code written in your language may look to you, it is YOUR language. When you look at it, it makes you cringe at your own reflection in the mirror. Have the will to respect yourself, because others don't care how much you love them instead of yourself. They won't appreciate it and will laugh at your attempts to emulate them. Worse is when the voluntary choice to be secondary is motivated by the fashion of making fun of yourself, and your own ventures are funded after confirming that others have successfully tried something similar.

Example:

The GPT-3 training (175 billion parameters) cost the Americans $26 million. (for a bank, this amount is a rubbish); the difference between GPT-3 and GPT-2 is only in size. That is, this technology has been around for a few years and OPEN AI was just scaling the network. So, could Sber (or someone) not have realized that this is a promising direction and invested $26m in a similar model first? I think they could have, but they didn't. Why didn't they? I think the decision to allocate money for similar projects is taken by an unspoken rule after the West succeeds in testing something similar and it becomes a standard. You understand that with this approach it is impossible to win the technological (and any) race, don't you? It turns out that in reality, you do not participate in any AI race, and just "roll on the tracks" laid by someone ahead and when they make a breakthrough, you find a confirmation of the reasons that you also need to accelerate, even though you know that you can not overtake anyhow, because you ride behind, like a "car" following the locomotive.

You are too fixated on other people's standards, and you fanatically elevate yours to "anti-standard", which is very pleasant and convenient for you know whom. Therein lies the impasse of your own development.

This is a subjecc tive opinion, possibly an emotional one.
 
Реter Konow:
I think it is because the decision to allocate money to such projects is made by an unspoken rule after the West has successfully tried something similar and it has become the standard.

That's a strange opinion. I don't think it is. We have ruined basic, fundamental science and lack of holistic learning. Pieces from ITMO, MEPhI do not solve problems with personnel, so there is no directional movement, and well-trodden paths are like that too.

And do not forget that our microelectronics is represented by only two processors Elbrus and Baikal, which are 15 years behind, while Baikal is also burdened with criminal prosecution. And there are no frames..... And China has also imposed restrictions on technology transfer. COCOM... was such an organisation.

It has been proven that development is fastest under stable market, or pseudo-market (the Chinese version) conditions. We do not have such conditions, and until they appear we will be catching up.

 
Valeriy Yastremskiy:

That's an odd opinion. I don't think it is. We have ruined basic, fundamental science and lack of holistic training. Pieces from ITMO, MEPhI do not close tasks with personnel, therefore there is no directional movement, and the beaten paths are well as the way too.

And do not forget that our microelectronics is represented by only two processors Elbrus and Baikal, which are 15 years behind, while Baikal is also burdened with criminal prosecution. And there are no frames..... And China has also imposed restrictions on technology transfer. COCOM... was such an organisation.

It has been proven that development is fastest under stable market, or pseudo-market (the Chinese version) conditions. We do not have such conditions, and until they appear we will be catching up.

But do we need it - to chase the elusive Joe, maybe wait for him to run and get up for a rest... and then some run fast, some crawl well and some fly fast - to whom what is given, technology emerges as needed and can take another revolutionary path...leapfrogging over a level... or in another level of thinking, like in another dimension or something... there's a magnetic cushion train in china, well it flies very fast, but it's expensive and therefore useless... so not all that gold... that runs fast, you have to live up to its value by a gradual movement to understand the need for it for everyone... like the underground - everyone needs it even though it's expensive and the speed will increase over time...

 
Valeriy Yastremskiy:

We have ruined basic, fundamental science and lack of holistic training. Pieces from ITMO, MEPhI do not cover the tasks with personnel, so there is no directional movement, and the beaten paths are well as a way too.

And do not forget that our microelectronics is represented by only 2 processors Elbrus and Baikal, which are 15 years behind, while Baikal is also burdened with criminal prosecution. And there are no cadres.....

Hands off those who once shut down Soviet computer projects in favour of copying IBM. And once well-known programmer (I don't remember exactly - Wirth or who, the old age is not a joy), whose multi-volume book on programming in 80th was by every Soviet programmer, especially studied Russian, to read the work of ak. Ershov.


Valeriy Yastremskiy:

COCOM... was such an organization.

COMCON... Massarakhsh-e-Massarakhsh!

 
Edgar Akhmadeev:

Hands off those who once shut down Soviet computer projects in favour of copying IBM. And once the famous programmer (I do not remember exactly - Wirth or who, old age is not a joy), whose multi-volume book on programming in the 80s was with every Soviet programmer, especially studied Russian to read the work of ac. Ershov.


COMCON... Massarakhsh-i-Massarakhsh!

I agree. The 360 was a critical error.

The Coordinating Committee for Export Controls, better known asCoCom orCOCOM, is an international organization established in 1949 to exercise multilateral export control over exports to the USSR and other socialist...
 
Accounting in Russia - clear, reliable and transparent https://startbusiness-russia.com/accounting
1...65666768697071727374
New comment
Reason: