Forumi


Povratak   PC Ekspert Forum > Ostalo > Svaštara
Ime
Lozinka

Odgovori
 
Uređivanje
Staro 06.04.2023., 22:12   #31
mann
Premium
 
mann's Avatar
 
Datum registracije: May 2004
Lokacija: Rijeka
Postovi: 2,151
ML, deep learning i ovo sto danas imamo sa fundation modelima ili large language modelima nije bas neka novost. Fundation model postoji vec jako dugo ali ok openai ga je jako popularizirao i priblizio masama. Razlike prvog, drugog i treceg je vise manje input koji je potreban da bi ovaj nesto naucio. Generalno sve je to jos uvijek statistika i model ali da je zgodno reci chatgptu da ti napise follow up email sa ovim zakljucima stvarno je. Brutalno on sroci taj tekst. Jedva cekam vidjeti njegov addon u wordu i outlooku direktno. Koji time saver ce to biti.

Sent using Tapatalk
mann je offline   Reply With Quote
Staro 07.04.2023., 09:26   #32
Inovator
Premium
Moj komp
 
Inovator's Avatar
 
Datum registracije: Nov 2005
Lokacija: Bjelovar
Postovi: 464
Sto mislite do koje mjere je dosao razvoj, ako je ovo sto je dato opcoj populaciji free s ovakvim mogucnostima...
Vjerojatno US army ima v10
Htjeli ne htjeli, stvar je tu i tesko da cemo nazad, pradavni strah covjeka je da ce ga zamijeniti stroj.
Inovator je offline   Reply With Quote
Staro 07.04.2023., 10:12   #33
mamutarka
ODUZIMAM VAM RIJEČ
Moj komp
 
mamutarka's Avatar
 
Datum registracije: Aug 2005
Lokacija: Zagreb
Postovi: 12,460
Citiraj:
Autor Inovator Pregled postova
Sto mislite do koje mjere je dosao razvoj
singularity
__________________
mamutarka je offline   Reply With Quote
Staro 08.04.2023., 11:41   #34
m4dm4n
Does youer dewg byte?
Moj komp
 
m4dm4n's Avatar
 
Datum registracije: May 2006
Lokacija: Zagreb
Postovi: 1,916
Zašto ljudi ovdje automatski diskvalificiraju AI kao neinteligenciju, samo zato jer nije AGI na razini IQ 500? Doslovno se za svaku životinju gleda što je sposobna u svom okolišu, kako se prilagođava, čak i mravi.
Imamo sustav koji na setu dostupnih naučenih podataka na temelju nekog ulaza uspije izgenerirati izlaz, te sam sebe ispravlja je li taj izlaz odgovarajuć. Ima hrpa tipova neuronskih mreža koje su odgovarajuće za pojedine tipove poslova, pokušavaju imitirati strukturu bioloških mreža.
Pa se zatim kaže, AI nema svijest o sebi, istina, no mnoge životinje nisu svjesne sebe, pa eto, pokazuju znakove inteligencije.

Eto, baš me zanima kako netko planira u budućnosti dokazati je li neki AI svjestan samog/samu sebe.

Sent from my SM-G991B using Tapatalk
__________________
" Have you tried turning it off and on again ? Are you sure it is plugged in ? "
m4dm4n je offline   Reply With Quote
Staro 08.04.2023., 12:21   #35
mamutarka
ODUZIMAM VAM RIJEČ
Moj komp
 
mamutarka's Avatar
 
Datum registracije: Aug 2005
Lokacija: Zagreb
Postovi: 12,460
__________________
mamutarka je offline   Reply With Quote
Staro 08.04.2023., 13:01   #36
OuttaControl
Premium
Moj komp
 
OuttaControl's Avatar
 
Datum registracije: Feb 2007
Lokacija: Dalmacija
Postovi: 5,216
Got schooled

OuttaControl je offline   Reply With Quote
Staro 08.04.2023., 13:02   #37
mamutarka
ODUZIMAM VAM RIJEČ
Moj komp
 
mamutarka's Avatar
 
Datum registracije: Aug 2005
Lokacija: Zagreb
Postovi: 12,460
ja nisam htio nastavit da me ne zbenaju
__________________
mamutarka je offline   Reply With Quote
Staro 08.04.2023., 13:52   #38
OuttaControl
Premium
Moj komp
 
OuttaControl's Avatar
 
Datum registracije: Feb 2007
Lokacija: Dalmacija
Postovi: 5,216
Nisam ni ja, doc berecki
OuttaControl je offline   Reply With Quote
Staro 08.04.2023., 13:53   #39
mamutarka
ODUZIMAM VAM RIJEČ
Moj komp
 
mamutarka's Avatar
 
Datum registracije: Aug 2005
Lokacija: Zagreb
Postovi: 12,460


zaš je prvi odgovor na eng a dalje na hr
__________________
mamutarka je offline   Reply With Quote
Staro 08.04.2023., 14:10   #40
Inovator
Premium
Moj komp
 
Inovator's Avatar
 
Datum registracije: Nov 2005
Lokacija: Bjelovar
Postovi: 464
ChatGPT je prosao i Turingov test
Inovator je offline   Reply With Quote
Staro 08.04.2023., 15:18   #41
OuttaControl
Premium
Moj komp
 
OuttaControl's Avatar
 
Datum registracije: Feb 2007
Lokacija: Dalmacija
Postovi: 5,216
Citiraj:
Autor mamutarka Pregled postova


zaš je prvi odgovor na eng a dalje na hr
Safety kicked off
OuttaControl je offline   Reply With Quote
Staro 08.04.2023., 15:35   #42
Bubba
E Pluribus UNIX
Moj komp
 
Bubba's Avatar
 
Datum registracije: Oct 2002
Lokacija: M82
Postovi: 6,544
Citiraj:
Autor Inovator Pregled postova
ChatGPT je prosao i Turingov test
"Turingov test" je arhaicni model koji je zamisljen kada je tranzistor bio opskurna i tesko dostupna tehnologija koja se svega 20ak godina prije toga smatrala tehnoloski neizvedivom.

Parafrazirajuci Clarka, "proci Turingov test" za elektronicki sklop je tada bila magija.

Citiraj:
Autor stitrek Pregled postova
Da, to je vezano uz komercijalni AI, pitanje je kakav AI Elon von Braun i ostatak ekipe imaju, ako to kroz 5-10 godina pocmu postepeno implementirati u mainstream, tada ce nastati velika zbrka oko određenih poslova, slicno kao i u industriji sa robotikom.
Kako god bilo, ljudi ce se snalaziti.
Citiraj:
Autor mkey Pregled postova
Vezano za gubitak poslova i priču o pripravnicima kazao bih slijedeće: jako puno ljudi iznimno loše rade svoj posao. Sporo ili nikako uče, ne vuku zaključke iz onoga što su napravili dobro ili loše, lijeni su i nezainteresirani te redovito slabo motivirani. Nemaju fokusa i koncentracija im je podijeljena na više fronti.
Doslovce ovo, da spojim citate.

Ako ti je posao ugrozen od moje Tesle P100 i malo dotreniranih LLaMA i Alpaca modela, onda imas sasvim drugi skup problema u zivotu.

Citiraj:
Autor Beren Pregled postova
Eliezera Yudkovskog, tipa koji zagovara polugodišnji moratorij na razvoj AI-ja i tvrdi da bi vlade trebale imati plan da po kratkom postupku srokaju podatkovne centre na kojem se AI vrti, ako zatreba i s nuklearnim oružjem.
Jel to brat burazer od onog Visageovog protezea, Google shizofrenika koji je "dokazao" da je neki tamo model svjestan sebe i da ga boli kada mu zbicim tri tokena ekstra?

Mozda da mu netko pokaze gdje je pancer na datacentru pa mu splasnu nuklearne ambicije.
__________________
Programer
Rok od dva mjeseca u stvari znači četiri, ali nikako ispod šest.
Bubba je offline   Reply With Quote
Staro 08.04.2023., 18:11   #43
kopija
DIY DILETANT
 
kopija's Avatar
 
Datum registracije: Jan 2009
Lokacija: Čistilište
Postovi: 3,060



$335,000 Pay for ‘AI Whisperer’ Jobs Appears in Red-Hot Market

Conrad Quilty-Harper
6-7 minutes


Everybody is talking about the artificial intelligence behind ChatGPT. Less noticed is a jobs market mushrooming around the technology, where these newly created roles can pay upwards of $335,000 a year.
And for many a computer engineering degree is optional.
They’re called “prompt engineers,” people who spend their day coaxing the AI to produce better results and help companies train their workforce to harness the tools.
Over a dozen artificial intelligence language systems called large language models, or LLMs, have been created by companies like Google parent Alphabet Inc., OpenAI and Meta Platforms Inc.. The technology has moved rapidly from experiments to practical use, with firms like Microsoft Corp. integrating ChatGPT into its Bing search engine and GitHub software development tool.
As the technology proliferates, many companies are finding they need someone to add rigor to their results.
“It’s like an AI whisperer,” says Albert Phelps, a prompt engineer at Mudano, part of consultancy firm Accenture in Leytonstone, England. “You’ll often find prompt engineers come from a history, philosophy, or English language background, because it’s wordplay. You're trying to distill the essence or meaning of something into a limited number of words.”
Phelps, 29, studied history at the University of Warwick near Birmingham, England, before starting his career as a consultant for banks like Clydesdale Bank and Barclays Plc, helping them solve problems around risk and regulations. A talk from the Alan Turing Institute, a UK-government funded institute for artificial intelligence, inspired him to research AI, leading to his role at Accenture.
He and colleagues spend most of the day writing messages or “prompts” for tools like OpenAI’s ChatGPT, which can be saved as presets within OpenAI’s playground for clients and others to use later. A typical day in the life of a prompt engineer involves writing five different prompts, with about 50 interactions with ChatGPT, says Phelps.
It’s too soon to know how widespread prompt engineering is or will become. The paradigm emerged in 2017 when AI researchers created “pre-trained” LLMs, which could be adapted to a wide range of tasks with the addition of a human text input. In the last year, LLMs like ChatGPT have attracted millions of users, who are all engaging in a form of prompt engineering whenever they tweak their prompts.


Anthropic is advertising for a “Prompt Engineer and Librarian” with a salary range of $175,000 to $335,000.
Credit: Anthropic
Companies like Anthropic, a Google-backed startup, are advertising salaries up to $335,000 for a “Prompt Engineer and Librarian” in San Francisco. Automated document reviewer Klarity also in California is offering as much as $230,000 for a machine learning engineer who can “prompt and understand how to produce the best output” from AI tools. Outside of the tech world, Boston Children’s Hospital and London law firm Mishcon de Reya recently advertised for prompt engineer jobs.
It’s now even possible to buy and sell text prompts via the PromptBase marketplace, which also helps people hire prompt engineers to create individual prompts for a fee.
The best-paying roles often go to people who have PhDs in machine learning or ethics, or those who have founded AI companies. Recruiters and others say these are among the critical skills needed to be successful.
“It’s probably the fastest-moving IT market I’ve worked in for 25 years,” says Mark Standen, who runs the staffing business for artificial intelligence, machine learning and automation at Hays in the UK and Ireland. “Salaries start at £40,000, but we’ve got candidates on our database looking for £200,000 to £300,000 a year. Expert prompt engineers can name their price.”
Google, TikTok and Netflix Inc. have been driving salaries higher, but the role is becoming mainstream among bigger companies thanks to the excitement around the launch of OpenAI’s ChatGPT-4, Google Bard and Microsoft’s Bing AI chatbot.
Outside of the tech world, companies in the financial, legal and insurance world are all experimenting with AI tools, which is also driving demand.
Mishcon de Reya recently advertised for a GPT Legal Prompt Engineer to help the firm understand how large language models could be applied to legal problems. The role is about finding out “how good are these models now? And how likely are they to meet any of the use cases?,” says Nick West, a partner and chief strategy officer at the firm. “I want to get on to the front edge of technology and play around with it.”
He cautions that paralegals looking for a step up in income might be disappointed. “We don’t need a £300,000 expert, that’s ludicrous money,” says West.
The industry’s warp-speed growth is already drawing comparisons to blockhain, Non-Fungible Tokens and crypto, which saw a jobs boom in 2021 with many firms offering hefty salaries and bonuses. Tech companies are now adding “AI” to their pitch decks to lure talent who have left or been forced out by jobs cuts in the crypto world. The jobs market may even be getting ahead of the technology.
“That might be a sign of froth,” says Tom Hewitson, founder at labworks.io, a conversation design studio which makes voice-controlled skills for Amazon Alexa in London. “The people best suited to doing this are product designers or business analysts comfortable working with AI who tend to earn £100,000 to £150,000.”
So anyone interested better move fast.
Prompts “are the main way people are interacting with these tools, and therefore being good at that is of high value,” says Adrian Weller, a director of research in machine learning at the University of Cambridge in Cambridge, England. “I wouldn’t be so sure that it will continue for a long time. Don’t dwell too much on the current state of prompt engineering. It’s starting to evolve quite quickly.”



-->
Dada, probaj se prijavit, traže šaptača AI-u.



$335,000 Pay for ‘AI Whisperer’ Jobs Appears in Red-Hot Market

Conrad Quilty-Harper
6-7 minutes


Everybody is talking about the artificial intelligence behind ChatGPT. Less noticed is a jobs market mushrooming around the technology, where these newly created roles can pay upwards of $335,000 a year.
And for many a computer engineering degree is optional.
They’re called “prompt engineers,” people who spend their day coaxing the AI to produce better results and help companies train their workforce to harness the tools.
Over a dozen artificial intelligence language systems called large language models, or LLMs, have been created by companies like Google parent Alphabet Inc., OpenAI and Meta Platforms Inc.. The technology has moved rapidly from experiments to practical use, with firms like Microsoft Corp. integrating ChatGPT into its Bing search engine and GitHub software development tool.
As the technology proliferates, many companies are finding they need someone to add rigor to their results.
“It’s like an AI whisperer,” says Albert Phelps, a prompt engineer at Mudano, part of consultancy firm Accenture in Leytonstone, England. “You’ll often find prompt engineers come from a history, philosophy, or English language background, because it’s wordplay. You're trying to distill the essence or meaning of something into a limited number of words.”
Phelps, 29, studied history at the University of Warwick near Birmingham, England, before starting his career as a consultant for banks like Clydesdale Bank and Barclays Plc, helping them solve problems around risk and regulations. A talk from the Alan Turing Institute, a UK-government funded institute for artificial intelligence, inspired him to research AI, leading to his role at Accenture.
He and colleagues spend most of the day writing messages or “prompts” for tools like OpenAI’s ChatGPT, which can be saved as presets within OpenAI’s playground for clients and others to use later. A typical day in the life of a prompt engineer involves writing five different prompts, with about 50 interactions with ChatGPT, says Phelps.
It’s too soon to know how widespread prompt engineering is or will become. The paradigm emerged in 2017 when AI researchers created “pre-trained” LLMs, which could be adapted to a wide range of tasks with the addition of a human text input. In the last year, LLMs like ChatGPT have attracted millions of users, who are all engaging in a form of prompt engineering whenever they tweak their prompts.


Anthropic is advertising for a “Prompt Engineer and Librarian” with a salary range of $175,000 to $335,000.
Credit: Anthropic
Companies like Anthropic, a Google-backed startup, are advertising salaries up to $335,000 for a “Prompt Engineer and Librarian” in San Francisco. Automated document reviewer Klarity also in California is offering as much as $230,000 for a machine learning engineer who can “prompt and understand how to produce the best output” from AI tools. Outside of the tech world, Boston Children’s Hospital and London law firm Mishcon de Reya recently advertised for prompt engineer jobs.
It’s now even possible to buy and sell text prompts via the PromptBase marketplace, which also helps people hire prompt engineers to create individual prompts for a fee.
The best-paying roles often go to people who have PhDs in machine learning or ethics, or those who have founded AI companies. Recruiters and others say these are among the critical skills needed to be successful.
“It’s probably the fastest-moving IT market I’ve worked in for 25 years,” says Mark Standen, who runs the staffing business for artificial intelligence, machine learning and automation at Hays in the UK and Ireland. “Salaries start at £40,000, but we’ve got candidates on our database looking for £200,000 to £300,000 a year. Expert prompt engineers can name their price.”
Google, TikTok and Netflix Inc. have been driving salaries higher, but the role is becoming mainstream among bigger companies thanks to the excitement around the launch of OpenAI’s ChatGPT-4, Google Bard and Microsoft’s Bing AI chatbot.
Outside of the tech world, companies in the financial, legal and insurance world are all experimenting with AI tools, which is also driving demand.
Mishcon de Reya recently advertised for a GPT Legal Prompt Engineer to help the firm understand how large language models could be applied to legal problems. The role is about finding out “how good are these models now? And how likely are they to meet any of the use cases?,” says Nick West, a partner and chief strategy officer at the firm. “I want to get on to the front edge of technology and play around with it.”
He cautions that paralegals looking for a step up in income might be disappointed. “We don’t need a £300,000 expert, that’s ludicrous money,” says West.
The industry’s warp-speed growth is already drawing comparisons to blockhain, Non-Fungible Tokens and crypto, which saw a jobs boom in 2021 with many firms offering hefty salaries and bonuses. Tech companies are now adding “AI” to their pitch decks to lure talent who have left or been forced out by jobs cuts in the crypto world. The jobs market may even be getting ahead of the technology.
“That might be a sign of froth,” says Tom Hewitson, founder at labworks.io, a conversation design studio which makes voice-controlled skills for Amazon Alexa in London. “The people best suited to doing this are product designers or business analysts comfortable working with AI who tend to earn £100,000 to £150,000.”
So anyone interested better move fast.
Prompts “are the main way people are interacting with these tools, and therefore being good at that is of high value,” says Adrian Weller, a director of research in machine learning at the University of Cambridge in Cambridge, England. “I wouldn’t be so sure that it will continue for a long time. Don’t dwell too much on the current state of prompt engineering. It’s starting to evolve quite quickly.”



kopija je offline   Reply With Quote
Staro 08.04.2023., 18:15   #44
medo
#erase startup-config
Moj komp
 
medo's Avatar
 
Datum registracije: Nov 2001
Lokacija: Zagreb
Postovi: 2,949
Umjetna inteligencija (AI)

Ako netko misli da u nadolazećim godinama AI neće promijeniti naše živote više nego mobiteli i internet zajedno taj je jako naivan.

Kao i sve koristit će se za dobre stvari kao i za loše stvari. Jedno je sigurno: ništa više neće biti isto.

Netko će se prilagoditi, netko neće. Mene je osobno najviše strah toga da ćemo funkciju razmišljanja, logike i strategije offloadati s naših mozgova na AI. Film “Idiocracy” bi za generaciju-dvije iza naših mogao u tom scenariju mogao izgledati kao sastanak članova Mense.
__________________
"It's not a bug, it's a feature!"
1N6pJsvusP7afu23qs1uBscK16wfcG7C8m
medo je offline   Reply With Quote
Staro 08.04.2023., 18:52   #45
The Exiled
McG
Moj komp
 
The Exiled's Avatar
 
Datum registracije: Feb 2014
Lokacija: Varaždin
Postovi: 6,772
IMHO nismo ni već danas predaleko od Idiocracy dokumentarca, posebice kad se pogleda kakav (negativan) efekt su u posljednjih 15 godina imale društvene mreže i "smart" nešto uređaji.

Hrpa zombificiranih neznalica luta uokolo, a kad im se još OpenAI, ChatGPT i općenito umjetna inteligencija stavi na raspolaganje - to je to - Don't Look Up u punom smislu tih riječi.
__________________
AMD Ryzen 7 Pro 4750G + Vega iGPU | be quiet! Pure Rock 2 Black | MSI B450 Tomahawk Max II | 32GB G.Skill DDR4-2666 Value | 256GB AData SX8200 Pro NVMe | 2x4TB WD Red Plus | Fractal Define 7 Compact | Corsair CX450M
The Exiled je offline   Reply With Quote
Staro 08.04.2023., 19:19   #46
Exy
Premium
Moj komp
 
Exy's Avatar
 
Datum registracije: Sep 2006
Lokacija: Zagreb, Črnomerec
Postovi: 1,981
Citiraj:
Autor dada-as Pregled postova
Da se vratimo u malo neozbiljne vode. Ovo je odgovor GPT-4 na post od kolege Exy. mene ovakvi odgovori plase... Od kud njemu te informacije i kak ih je kvragu natipkao za par sekundi bez da je prvo sjeo i malo razmislio.
Što misliš odakle mu te informacije? Mislim, volio bih da sam kao Albert Einstein pa da mogu pisati članke kojima izvori i fusnote ne trebaju, zato što njihov sadržaj nikad prije nije spomenut ali nažalost sve što sam napisao postoji negdje u bespućima milijarda članaka kojima chatgpt ima pristup. Pa tako naravno i argumentima, primjerima itd.
Iz mog iskustva, nečije udivljenje chatgpt-om je obrnuto proporcionalno količini znanja o tome kako ta stvar uopće radi.
Daleko od toga da tvrdim kako mi je to 100% poznato, ali neku opću ideju o tome ipak imam.
Na isti način kao što google odgovara na moje upute, to isto čini i chatgpt. Za razliku od google-a on to formatira u odgovarajući tekst ali dublje poznavanje konteksta o čemu se govori naravno ne postoji.
Suhoparno rečeno, chatgpt proizvodi tekst koji statistički odgovara upitu koji mu je postavljen, i to je to. Ako to znate i razumijete što to znači, već ste ispred 99% ljudi na ovom svijetu kao i ispred 100% novinara
Chatgpt ništa "ne zna", on se bavi mapiranje govornih obrazaca. To radi vrlo vrlo dobro zbog suludo velike procesne snage i suludo velike baze kojoj ima pristup. No kvantiteta ne proizvodi kvalitetu. To ne znači da to nije alat koji može biti izuzetno koristan, naprotiv.

Pitanja tipa hoće li se moj sir rastopiti u frižideru ili koja je treća riječ ove rečenice naravno zbunjuju ubogu skripticu, iako ne sumnjam da će ekipa od 50-ak ljudi koja ju svakodnevno hrani podacima i trenira sa vremenom ubaciti lijek za većinu tih ahilovih peta. Na sreću neznalica i novinara širom svijeta

Što se Turinga tiče, tu bi bilo korisno baciti oko na njegov originalni rad, onda čovjek lako vidi da su vijesti o tome kako je chatgpt prošao turinga opet novinarski senzacionalizam. Ne kažem da je nemoguće da on to napravi, ali to bi značilo vrlo malo, pogotovo što o tome da li je stroj prošao ili ne odlučuje čovjek. Sjećam se jednog predavanja Jarona Laniera gdje je spomenuo kako u razvoju sve "pametnijeg" AI-a nismo računali na to da ljudi postaju sve gluplji
Turing je arhaična stvarčica i većina ljudi iz polja AI-a (stručnjaka, ne prodavača magle tipa Ray Kurzweil) se slažu da bi ga trebalo apdejtati kako shvaćamo sve više ne o računalima nego o ljudskom mozgu. Inače, kao ni za vrijeme Turinga, tako ni danas i dalje nemamo teoriju o tome kako mozak uopće radi.

No sve u svemu, riječ je o genijalnom komadu tehnologije, potencijalno izuzetno disruptivnom (u dobrom i lošem smislu) za društvo, ali otprilike jednako inteligentnom, svjesnom, pametnom i osjetljivom kao što je to moj printer.

Zadnje izmijenjeno od: Exy. 08.04.2023. u 19:24.
Exy je offline   Reply With Quote
Staro 08.04.2023., 19:45   #47
mkey
Premium
Moj komp
 
Datum registracije: Sep 2018
Lokacija: tu
Postovi: 1,937
Citiraj:
Autor mann Pregled postova
ML, deep learning i ovo sto danas imamo sa fundation modelima ili large language modelima nije bas neka novost. Fundation model postoji vec jako dugo ali ok openai ga je jako popularizirao i priblizio masama. Razlike prvog, drugog i treceg je vise manje input koji je potreban da bi ovaj nesto naucio. Generalno sve je to jos uvijek statistika i model ali da je zgodno reci chatgptu da ti napise follow up email sa ovim zakljucima stvarno je. Brutalno on sroci taj tekst. Jedva cekam vidjeti njegov addon u wordu i outlooku direktno. Koji time saver ce to biti.

Sent using Tapatalk
Chatbot će generirati mail kojeg ćeš ti poslalti i mail kojeg će tebi poslati kao odgovor na isti? Ne znam tko je bot u transakciji gdje se chatbot dopisuje preko dva čovjeka.
mkey je offline   Reply With Quote
Staro 08.04.2023., 21:34   #48
medo
#erase startup-config
Moj komp
 
medo's Avatar
 
Datum registracije: Nov 2001
Lokacija: Zagreb
Postovi: 2,949
Umjetna inteligencija (AI)