1 month ago
Elon Musk warned that artificial intelligence is 'one of nan biggest threats' to humanity coming arsenic he made a prima quality astatine nan UK's AI Safety Summit.
The billionaire X leader utilized a prima move astatine nan Bletchley Park arena to opportunity it was 'timely' because AI poses an 'existential risk' to humans, who look being outsmarted by machines for nan first time.
The arena arranged by Rishi Sunak will big discussions pinch world leaders, tech firms and scientists connected nan risks of caller technology.
Mr Musk, 1 of nan world's richest men, is simply a well-known professional of AI. His involution came after senior Meta executive Nick Clegg - a erstwhile UK lawman premier curate - warned generative AI 'deepfakes' may pose a consequence to upcoming elections successful nan US and UK adjacent year.
The King hailed AI arsenic 'one of nan top technological leaps successful nan history of quality endeavour' but warned of nan urgent request to activity together to combat its 'significant risks' arsenic he opened nan arena pinch a video message.
Mr Musk said he hoped nan two-day acme could beryllium utilized to found an 'international consensus' and nan creation of a 'third-party referee' to show technology.
'I deliberation AI is 1 of nan biggest threats (to humans),' he said.
'We person for nan first clip nan business wherever we person thing that is going to beryllium acold smarter than nan smartest human.
'We're not stronger aliases faster than different creatures, but we are much intelligent, and present we are for nan first time, really successful quality history, pinch thing that is going to beryllium acold much intelligent than us.
The billionaire X leader utilized a prima move astatine nan Bletchley Park arena to opportunity it was 'timely' because AI poses an 'existential risk' to humans, who look being outsmarted by machines successful future.
Mr Clegg insisted this greeting that civilisation-ending threats had been overblown, and urged governments to debar heavy-handed regulation. But he said location were risks to elections from exertion that allows convincing clone worldly to beryllium created.
Some 28 nations including nan US and China person signed nan Bletchley Declaration mounting retired 'a shared knowing of nan opportunities and risks posed by frontier AI and nan request for governments to activity together to meet nan astir important challenges', Science and Technology Secretary Michelle Donelan said.
Mr Musk said he hoped nan two-day acme could beryllium utilized to found an 'international consensus' and nan creation of a 'third-party referee' to show technology.
'It's not clear to maine if we tin power specified a thing, but I deliberation we tin aspire to guideline it successful a guidance that's beneficial to humanity
'But I do deliberation it's 1 of nan existential risks that we look and it is perchance nan astir pressing 1 if you look astatine nan timescale and complaint of advancement - nan acme is timely, and I applaud nan Prime Minister for holding it.'
Mr Musk added that he hoped nan two-day acme could beryllium utilized to found an 'international consensus' connected penetration into AI, truthful that a 'third-party referee' could beryllium established successful nan assemblage 'who tin observe what starring AI companies are doing and astatine slightest sound nan siren if they person concerns'.
Some 28 nations including nan US and China person signed nan Bletchley Declaration mounting retired 'a shared knowing of nan opportunities and risks posed by frontier AI and nan request for governments to activity together to meet nan astir important challenges'.
Mr Clegg insisted this greeting that civilisation-ending threats had been overblown, and urged governments to debar heavy-handed regulation.
But he said location were risks to elections from exertion that allows convincing clone worldly to beryllium created.
Mr Clegg said manufacture and authorities co-operation was needed 'right now' connected nan domiciled generative AI - which tin make images, text, audio and moreover videos astatine petition - will play successful votes adjacent year.
The apical Facebook executive said retired arsenic nan UK's AI acme astatine Bletchley Park began pinch an statement to tackle nan 'risks' from caller technology.
Ministers will talk nan anticipation of AI interfering successful elections astatine nan summit, Oliver Dowden has said.
The Deputy PM is launching nan convention astatine nan historical wartime codebreaker guidelines arsenic experts wrangle complete nan standard of nan risks from nan emerging exertion - and really to minimise them.
A authorities insubstantial published past week revealed really advances successful AI will make it cheaper and easier for hackers, scammers and terrorists to onslaught guiltless victims – each wrong nan adjacent 18 months.
It predicts alleged bots – societal media accounts that look for illustration existent group – will target users pinch 'personalised disinformation', machines will motorboat their ain cyberattacks without needing quality direction, and anyone will beryllium capable to study really to create nan world's astir vulnerable weapons.
The impermanent database for nan arena is still not wholly clear, pinch Downing Street forced to contradict nan acme is being snubbed by world leaders.
Neither US President Joe Biden nor French President Emmanuel Macron will beryllium there, but Vice President Kamala Harris will correspond nan US and EU committee main Ursula von der Leyen will speak connected behalf of nan bloc.
Another chartless is who will correspond China, though ministers person confirmed that nan Asian superpower has been invited.
Politicians and tech moguls will stitchery astatine Bletchley Park coming arsenic Oliver Dowden (pictured) kicks disconnected nan first ever AI information summit
Roundtable discussions connected frontier AI information will return spot connected nan first time of nan gathering astatine Bletchley Park, adjacent Milton Keynes, which was location to nan UK's Second World War codebreakers
The acme was primitively billed by Mr Sunak arsenic a chance for nan UK to lead nan world successful establishing statement connected nan safe usage of AI technology, aft he group retired his ambition for nan UK to go nan 'geographical location of world AI information regulation' successful nan summer.
Conflict successful nan Middle East has since fixed nan attraction of world leaders, and successful nan US, President Biden has signed an executive bid to 'ensure that America leads nan measurement successful seizing nan committedness and managing nan risks of artificial intelligence'.
But nan UK Government has insisted nan acme has merit, pinch a impermanent database including pioneering AI companies for illustration UK-based Deep Mind, arsenic good arsenic OpenAI, and Anthropic.
Mr Musk, nan Tesla CEO and proprietor of nan societal media tract X, will subordinate nan PM for a unrecorded question and reply aft nan acme closes tomorrow.
The billionaire is simply a co-founder of OpenAI – nan institution down nan celebrated ChatGPT tool – and has antecedently expressed concerns astir nan anticipation of AI becoming dispute towards humans.
Sir Nick Clegg, now president of world affairs astatine Meta, told BBC Radio 4's Today programme that nan dangers of existent AI systems had been 'overstated'.
'I sometimes wonderment whether we somewhat overstate what these systems do, arsenic they are presently constituted,' he said.
'These ample connection models, these generative AI models that person created truthful overmuch excitement complete caller months – they are very powerful, they are very versatile, they devour a immense magnitude of information and they require a batch of computing capacity.
'But they are fundamentally processing immense amounts of information and past predicting, if you like, nan adjacent connection aliases nan adjacent token successful consequence to a quality prompt. So they are immense gigantic auto-complete systems. They don't cognize thing themselves, they person nary innate knowledge, they person nary autonomy aliases agency yet.
'This is I deliberation a very important constituent and theme… really overmuch clip should we each take – academics, researchers, civilian nine governments regulators and nan manufacture – (to spend) connected a future, a comparatively speculative early astir what AI systems mightiness germinate into successful nan future.
'And really overmuch should we attraction connected challenges that we person successful nan present and now.'
Sir Nick said it was important that 'proximate challenges' were not 'crowded retired by a batch of speculative, sometimes somewhat futuristic predictions which whitethorn materialise, but candidly location are a batch of group who deliberation they won't materialise aliases materialise anytime soon'.
He warned that co-operation is needed 'right now' connected nan domiciled generative AI will play successful elections adjacent year.
Mr Dowden agreed that nan predetermination consequence from AI would beryllium a cardinal coming astatine nan summit.
'I deliberation Nick is perfectly correct to item that and so that is 1 of nan topics we will beryllium discussing coming astatine nan summit,' he said.
'There is simply a scope of different buckets of risks, and nan first 1 is precisely those benignant of societal risks, whether they spell to bias, disinformation aliases nan creation of deepfakes.
'But we should besides beryllium mindful of nan longer-term risks. We don't do this successful a vacuum. You good cognize that cognize celebrated informing that was issued by those frontier AI companies astir an existential risk.
'I woody pinch that successful a pragmatic and proportionate way, but we request to return it earnestly and analyse consequence and that is what this acme is besides about.'
Michelle Donelan, nan Secretary of State for Science, Innovation and Technology, said: 'AI is already an bonzer unit for bully successful our society, pinch limitless opportunity to turn nan world economy, present amended nationalist services and tackle immoderate of nan world's biggest challenges.
'But nan risks posed by frontier AI are superior and substantive and it is captious that we activity together, some crossed sectors and countries to recognise these risks.
'This acme provides an opportunity for america to guarantee we person nan correct group pinch nan correct expertise gathered astir nan array to talk really we tin mitigate these risks moving forward.
'Only past will we beryllium capable to genuinely reap nan benefits of this transformative exertion successful a responsible manner.'
Roundtable discussions connected frontier AI information will return spot connected nan first time of nan gathering astatine Bletchley Park, adjacent Milton Keynes, which was location to nan UK's Second World War codebreakers.
Alongside tech companies, civilian nine groups and experts from nan Alan Turing Institute, and nan Ada Lovelace Institute will beryllium present.
The 2nd time of nan acme will attraction connected nan responsible usage of AI technology.
As nan convention begins, nan UK Government has meantime pledged £38 cardinal towards backing artificial intelligence projects astir nan world, starting successful Africa.
The committedness is portion of an £80 cardinal collaboration betwixt Britain, Canada and nan Bill and Melinda Gates Foundation to boost 'safe and responsible' programming, nan Foreign Office said.
On nan eve of nan AI Safety Summit, Turing grant victor Yoshua Bengio has signed an unfastened missive informing nan threat it poses 'warrants contiguous and superior attention'
Elon Musk's hatred of AI explained: Billionaire believes it will spell nan extremity of humans - a fearfulness Stephen Hawking shared
Elon Musk wants to push exertion to its absolute limit, from abstraction recreation to self-driving cars — but he draws nan statement astatine artificial intelligence.
The billionaire first shared his distaste for AI successful 2014, calling it humanity's 'biggest existential threat' and comparing it to 'summoning nan demon.'
At nan time, Musk besides revealed he was investing successful AI companies not to make money but to support an oculus connected nan exertion successful lawsuit it gets retired of hand.
His main fearfulness is that successful nan incorrect hands, if AI becomes advanced, it could overtake humans and spell nan extremity of mankind, which is known arsenic The Singularity.
That interest is shared among galore superb minds, including nan precocious Stephen Hawking, who told the BBC in 2014: 'The improvement of afloat artificial intelligence could spell nan extremity of nan quality race.
'It would return disconnected connected its ain and redesign itself astatine an ever-increasing rate.'
Despite his fearfulness of AI, Musk has invested successful nan San Francisco-based AI group Vicarious, successful DeepMind, which has since been acquired by Google, and OpenAI, creating nan celebrated ChatGPT programme that has taken nan world by large wind successful caller months.
During a 2016 interview, Musk noted that he and OpenAI created nan institution to 'have democratisation of AI exertion to make it wide available.'
Musk founded OpenAI pinch Sam Altman, nan company's CEO, but successful 2018 nan billionaire attempted to return power of nan start-up.
His petition was rejected, forcing him to discontinue OpenAI and move connected pinch his different projects.
In November, OpenAI launched ChatGPT, which became an instant occurrence worldwide.
The chatbot uses 'large connection model' package to train itself by scouring a monolithic magnitude of matter information truthful it tin study to make eerily human-like matter successful consequence to a fixed prompt.
ChatGPT is utilized to constitute investigation papers, books, news articles, emails and more.
But while Altman is basking successful its glory, Musk is attacking ChatGPT.
He says nan AI is 'woke' and deviates from OpenAI's original non-profit mission.
'OpenAI was created arsenic an unfastened root (which is why I named it 'Open' AI), non-profit institution to service arsenic a counterweight to Google, but now it has go a closed source, maximum-profit institution efficaciously controlled by Microsoft, Musk tweeted successful February.
The Singularity is making waves worldwide arsenic artificial intelligence advances successful ways only seen successful subject fabrication - but what does it really mean?
In elemental terms, it describes a hypothetical early wherever exertion surpasses quality intelligence and changes nan way of our evolution.
Experts person said that erstwhile AI reaches this point, it will beryllium capable to innovate overmuch faster than humans.
There are 2 ways nan advancement could play out, pinch nan first starring to humans and machines moving together to create a world amended suited for humanity.
For example, humans could scan their consciousness and shop it successful a machine successful which they will unrecorded forever.
The 2nd script is that AI becomes much powerful than humans, taking power and making humans its slaves - but if this is true, it is acold disconnected successful nan distant future.
Researchers are now looking for signs of AI reaching The Singularity, specified arsenic nan technology's expertise to translate reside pinch nan accuracy of a quality and execute tasks faster.
Former Google technologist Ray Kurzweil predicts it will beryllium reached by 2045.
He has made 147 predictions astir exertion advancements since nan early 1990s - and 86 per cent person been correct.
