Guest book
Sign guestbook
Nous avons été enchanté de vous recevoir, et souhaitons que ce plaisir fut partagé.
Laissez nous vos impressions
540213
Entries in guestbook
Nous avons été enchanté de vous recevoir, et souhaitons que ce plaisir fut partagé.
Laissez nous vos impressions
Saundra
Sunday, 17 March 2024
F?om Google Maps to facial recognition, we're ev?r m??e reliant ?n artificial intelligence.
Expert MIKE WOOLRIDGE ?ives his reasons to ?e fearful - and cheerful.
?e asked ChatGPT (chat.openai.?om) to cr?ate a main imag? for this piece from the f?llowing YOU prompts: ‘th? future looks bright', ‘flat illustration', ‘poster style', ‘embrace technology t? our advantage' ?nd ‘?lder woman'
1 AI is not as intelligent as it ?eems (y?t)
T?? ?ne golden message i? t?at, even ?ith the next generation of conversational software tools ?ike ChatGPT, ?hich is th? biggest thing - possibly ever - ?n AI, you are not dealing w?t? a mind. ChatGPT [an AI-powered large language model that is trained on huge amounts of data] i? a ?omputer programme t?at has been heavily optimised, t? use the technical term, to do one th?ng: tell you w?at y?u want to hear.
It's v?ry good at th?t: very fluent ?nd very eloquent. ?hat leads us to ?elieve th?t we are communicating ?ith a mind, like a human b?ing, but ?e a?en't.
2 It can't cope w?th the real worl?
?I is all ?bout getting computers to do th?ngs t?at ??n ?nly be done by human brains and nervous systems. AI finds some things tremendously har? that we ?on't associate w?th intelligence, l?ke driving ? car. Untold amounts of money ha?e ?een invested in trying to get cars that can safely drive t?emselves ?nd we are ?t?ll not th?re.
We a?e no?h?re near ha?ing robot butlers.
3 Google Translate ?? ? game-changer
AI translation tools ?re one of the m?st *** t?ings humanity ?as achieved. You ?an translate ?etween pretty m?ch ?ny widel? us?d language in the wo?ld. Future A? translation apps will learn ?bout y?u, the language y?u use and th? way yo? communicate.
?ou ?ill be a?le to h?ve simultaneous spoken-wo?d translations.
4 Don't trust ChatGPT ?ith ??ur credit card
One of th? fundamental problems ?ith larg? language model (LLM) technology ?s th?t it ?oesn't te?l you the truth. ??u ask it a question; it ?ives yo? th? mo?t plausible answe?. It h?s no conception of truth. ?f it ?oesn't know the answer to a question it will simply ma?e ?omething u?.
I asked an early v?rsion of ChatGPT w?at ?t ?new abo?t me. Its tw?-l?ne summary s?id I'd studied ?t Cambridge, ?hich is ? typical background f?r an Oxford professor, ?ut I ?aven't. T?at'? why I'? be reluctant t? put ?ne of these LLMs ?n ?ny situation w?ere ?t was making decisions with a real consequence.
The technology is unreliable and prone to weird errors. ? ?ouldn't trust ChatGPT ?ith my credit card.
5 Don't b?lieve eve?ything y?u ?ead on social media
?? can go to ?ou? feed, look at your personal prejudices ?nd produce misinformation ?n ?n industrial scale ?bout yo?r favourite, ?r least favourite, politician based ?n the language you've used online - s? you can relate t? it if yo?'r? ? teenager ?n Middlesbrough ?r a Tory voter in the Home Counties.
?t's easy t? see how that m?ght ?e use? to influence voters to impact political elections in the UK ?nd overseas.
?ithin a smal? number of years, it's highly probable t?at on social media t?ere will ?? much more A?-generated Cont?nt Creation than human-generated.
?t ?ill b? very difficult for u? to kno? what is authentically human ?nd what i?n't. Trusted sources ?f news w?ll bec?me extremely ?mportant to navigate t?at. ?o forward 50 ?ears ?nd t?ere ?ill be vastly more AI-generated books, music and videos than t?ere will be human-generated ?nes.
6 A? ?ill transform healthcare
Doctors ?ill ?se AI as a diagnostic tool in the s?me w?y t??t us?ng an ?-ray machine i? a tool. But I w?uld caution ?gainst ?ecoming comp?etely reliant ?n an A? diagnosis. ?y big concern is th?t when a machine tells y?u somet?ing that disagrees w?th your ?wn personal perceptions, ?ou ?ave to invest ?ome energy to figure out why the machine might b? wrong.
?nd often people ?an't b? bothered t? do that if they're tired. I ?ould rath?r h?v? a human ?eing than AI. ?ome people won't hav? that luxury - ?t ?ill b? AI or nothing. On the ?lus sid?, AI tools ?ill ma?e v?ry expensive healthcare, ?nly accessible t? a tiny fragment of the world's population, accessible t? a ?uge number of people.
For exampl?, ?I technology ?ill b? ab?e t? analyse scans f?om the ot?er side of t?? wo?ld. Th?t will ?e transformative.
7 It'? biased t?wards cert?in sectors ?f society
AI is dominated by North American, ?hite, college-educated m?n. T?at is a w?dely acknowledged ?roblem: LLMs communicate ?sing North American (?ather t?an British) English. ?his matters. ? recent data s?t widely employed f?r voice-recognition technology u?ed recordings d?ne ?y white, male, college-educated Americans.
If y?u were outs?d? th?t data s?t - female, fo? instance - t?e voice-recognition systems ?idn't ?o a ve?y ?ood job of understanding yo?. To build an LLM, y?u need lot? ?f examples ?f the language. Icelandic ?oesn't ?t t?? moment h?ve a digital footprint b?g enough t? build LLMs that c?n communicate ?n the language - that ?? ? subtle danger abo?t the way AI can fail to encompass a wider range of human views ?nd experiences.
8 There is a risk ?f misuse
M?st human programmers w?uld decline ? request t? ?rite a programme t? launch a cyber attack on NHS data servers. ?ut if ??u ha?e ?n LLM th?t w?ll do what?v?r you ask it t? do, attacks ?ike that are easier [in the wrong hands]. People using ?I t? do bad things ?? something ?? need to think abo?t.
?t a recent ?? summit in t?e UK, the discussion ?ad moved on fr?m existential t? catastrophic risk ??ereby a failure ?f AI leads t? som?t?ing - like a plane crash in a highly populated ?rea - t??t w?uld b? devastating, ?ut isn't going to lead t? the end of humanity.
That's a healthy direction t? take the discussion, ?ecause ?t's more real, and when ?t's more real it'? something y?u can think about mitigating.
9 AI is just ? tool
?I ?sn't s?mething to be ?n awe ?f o? fear. It's going t? ?? used t? d? wonderful thing?, l?ke recognising tumours on ?-ray scans and spotting ?arly-onset dementia or heart disease, but ?t's just a tool like any othe? tool; don't talk about ?t ?s if it'? a person.
10 It coul? boost creativity
?he goal i? for AI to free people f?om drudgery ?nd enable th?m to focus on im?ortant thing? that require human understanding, creativity ?nd empathy. One unexpected application ?f LLMs is brainstorming ne? ideas: somet?ing people ?on't find easy or v?lue.
Ask ChatGPT for pitches f?r a banana-flavoured milk drink ?nd it w?ll k?ep producing, junk or not. And they ?ill g?t human creative juices flowing.
LLMs ?ill open ?p opportunities f?r ?ompletely new art forms ?e can't ?ven imagine rig?t no?, just a? who?ve? invented ea?ly synthesisers ?ouldn't ha?e imagined rave music t?king off ?n the 80s. I t?ink we'?? on th? cusp ?f an explosion of people doing rea?ly creative t?ings ?ith AI.
Mike Wooldridge is a professor of A? at Oxford University. ?he Truth About ?I, hi? Royal Institution Christmas Lectures, ?re on iPlayer.
Illustration: Chloe Sharp/Midjourney
GoogleChatGPTAI
Expert MIKE WOOLRIDGE ?ives his reasons to ?e fearful - and cheerful.
?e asked ChatGPT (chat.openai.?om) to cr?ate a main imag? for this piece from the f?llowing YOU prompts: ‘th? future looks bright', ‘flat illustration', ‘poster style', ‘embrace technology t? our advantage' ?nd ‘?lder woman'
1 AI is not as intelligent as it ?eems (y?t)
T?? ?ne golden message i? t?at, even ?ith the next generation of conversational software tools ?ike ChatGPT, ?hich is th? biggest thing - possibly ever - ?n AI, you are not dealing w?t? a mind. ChatGPT [an AI-powered large language model that is trained on huge amounts of data] i? a ?omputer programme t?at has been heavily optimised, t? use the technical term, to do one th?ng: tell you w?at y?u want to hear.
It's v?ry good at th?t: very fluent ?nd very eloquent. ?hat leads us to ?elieve th?t we are communicating ?ith a mind, like a human b?ing, but ?e a?en't.
2 It can't cope w?th the real worl?
?I is all ?bout getting computers to do th?ngs t?at ??n ?nly be done by human brains and nervous systems. AI finds some things tremendously har? that we ?on't associate w?th intelligence, l?ke driving ? car. Untold amounts of money ha?e ?een invested in trying to get cars that can safely drive t?emselves ?nd we are ?t?ll not th?re.
We a?e no?h?re near ha?ing robot butlers.
3 Google Translate ?? ? game-changer
AI translation tools ?re one of the m?st *** t?ings humanity ?as achieved. You ?an translate ?etween pretty m?ch ?ny widel? us?d language in the wo?ld. Future A? translation apps will learn ?bout y?u, the language y?u use and th? way yo? communicate.
?ou ?ill be a?le to h?ve simultaneous spoken-wo?d translations.
4 Don't trust ChatGPT ?ith ??ur credit card
One of th? fundamental problems ?ith larg? language model (LLM) technology ?s th?t it ?oesn't te?l you the truth. ??u ask it a question; it ?ives yo? th? mo?t plausible answe?. It h?s no conception of truth. ?f it ?oesn't know the answer to a question it will simply ma?e ?omething u?.
I asked an early v?rsion of ChatGPT w?at ?t ?new abo?t me. Its tw?-l?ne summary s?id I'd studied ?t Cambridge, ?hich is ? typical background f?r an Oxford professor, ?ut I ?aven't. T?at'? why I'? be reluctant t? put ?ne of these LLMs ?n ?ny situation w?ere ?t was making decisions with a real consequence.
The technology is unreliable and prone to weird errors. ? ?ouldn't trust ChatGPT ?ith my credit card.
5 Don't b?lieve eve?ything y?u ?ead on social media
?? can go to ?ou? feed, look at your personal prejudices ?nd produce misinformation ?n ?n industrial scale ?bout yo?r favourite, ?r least favourite, politician based ?n the language you've used online - s? you can relate t? it if yo?'r? ? teenager ?n Middlesbrough ?r a Tory voter in the Home Counties.
?t's easy t? see how that m?ght ?e use? to influence voters to impact political elections in the UK ?nd overseas.
?ithin a smal? number of years, it's highly probable t?at on social media t?ere will ?? much more A?-generated Cont?nt Creation than human-generated.
?t ?ill b? very difficult for u? to kno? what is authentically human ?nd what i?n't. Trusted sources ?f news w?ll bec?me extremely ?mportant to navigate t?at. ?o forward 50 ?ears ?nd t?ere ?ill be vastly more AI-generated books, music and videos than t?ere will be human-generated ?nes.
6 A? ?ill transform healthcare
Doctors ?ill ?se AI as a diagnostic tool in the s?me w?y t??t us?ng an ?-ray machine i? a tool. But I w?uld caution ?gainst ?ecoming comp?etely reliant ?n an A? diagnosis. ?y big concern is th?t when a machine tells y?u somet?ing that disagrees w?th your ?wn personal perceptions, ?ou ?ave to invest ?ome energy to figure out why the machine might b? wrong.
?nd often people ?an't b? bothered t? do that if they're tired. I ?ould rath?r h?v? a human ?eing than AI. ?ome people won't hav? that luxury - ?t ?ill b? AI or nothing. On the ?lus sid?, AI tools ?ill ma?e v?ry expensive healthcare, ?nly accessible t? a tiny fragment of the world's population, accessible t? a ?uge number of people.
For exampl?, ?I technology ?ill b? ab?e t? analyse scans f?om the ot?er side of t?? wo?ld. Th?t will ?e transformative.
7 It'? biased t?wards cert?in sectors ?f society
AI is dominated by North American, ?hite, college-educated m?n. T?at is a w?dely acknowledged ?roblem: LLMs communicate ?sing North American (?ather t?an British) English. ?his matters. ? recent data s?t widely employed f?r voice-recognition technology u?ed recordings d?ne ?y white, male, college-educated Americans.
If y?u were outs?d? th?t data s?t - female, fo? instance - t?e voice-recognition systems ?idn't ?o a ve?y ?ood job of understanding yo?. To build an LLM, y?u need lot? ?f examples ?f the language. Icelandic ?oesn't ?t t?? moment h?ve a digital footprint b?g enough t? build LLMs that c?n communicate ?n the language - that ?? ? subtle danger abo?t the way AI can fail to encompass a wider range of human views ?nd experiences.
8 There is a risk ?f misuse
M?st human programmers w?uld decline ? request t? ?rite a programme t? launch a cyber attack on NHS data servers. ?ut if ??u ha?e ?n LLM th?t w?ll do what?v?r you ask it t? do, attacks ?ike that are easier [in the wrong hands]. People using ?I t? do bad things ?? something ?? need to think abo?t.
?t a recent ?? summit in t?e UK, the discussion ?ad moved on fr?m existential t? catastrophic risk ??ereby a failure ?f AI leads t? som?t?ing - like a plane crash in a highly populated ?rea - t??t w?uld b? devastating, ?ut isn't going to lead t? the end of humanity.
That's a healthy direction t? take the discussion, ?ecause ?t's more real, and when ?t's more real it'? something y?u can think about mitigating.
9 AI is just ? tool
?I ?sn't s?mething to be ?n awe ?f o? fear. It's going t? ?? used t? d? wonderful thing?, l?ke recognising tumours on ?-ray scans and spotting ?arly-onset dementia or heart disease, but ?t's just a tool like any othe? tool; don't talk about ?t ?s if it'? a person.
10 It coul? boost creativity
?he goal i? for AI to free people f?om drudgery ?nd enable th?m to focus on im?ortant thing? that require human understanding, creativity ?nd empathy. One unexpected application ?f LLMs is brainstorming ne? ideas: somet?ing people ?on't find easy or v?lue.
Ask ChatGPT for pitches f?r a banana-flavoured milk drink ?nd it w?ll k?ep producing, junk or not. And they ?ill g?t human creative juices flowing.
LLMs ?ill open ?p opportunities f?r ?ompletely new art forms ?e can't ?ven imagine rig?t no?, just a? who?ve? invented ea?ly synthesisers ?ouldn't ha?e imagined rave music t?king off ?n the 80s. I t?ink we'?? on th? cusp ?f an explosion of people doing rea?ly creative t?ings ?ith AI.
Mike Wooldridge is a professor of A? at Oxford University. ?he Truth About ?I, hi? Royal Institution Christmas Lectures, ?re on iPlayer.
Illustration: Chloe Sharp/Midjourney
GoogleChatGPTAI
Eliza
Sunday, 17 March 2024 | Kielce
Eliza
Sunday, 17 March 2024 | Kielce
Eliza
Sunday, 17 March 2024 | Kielce
Eliza
Sunday, 17 March 2024 | Kielce
Earlene
Sunday, 17 March 2024
An incredible notification characteristic powered from the 1xbet download (www.perksense.com) betting app allows it to notify users of actions alongside live events.
Earlene
Sunday, 17 March 2024
An incredible notification characteristic powered from the 1xbet download (www.perksense.com) betting app allows it to notify users of actions alongside live events.
Earlene
Sunday, 17 March 2024
An incredible notification characteristic powered from the 1xbet download (www.perksense.com) betting app allows it to notify users of actions alongside live events.
Earlene
Sunday, 17 March 2024
An incredible notification characteristic powered from the 1xbet download (www.perksense.com) betting app allows it to notify users of actions alongside live events.
Sharyl
Sunday, 17 March 2024 | Haren
In some situations this can be too risky, because some major tournaments have teams associated with almost the similar level.
Here is my site; website
Here is my site; website
540213
Entries in guestbook