Could ChatGPT REALLY slay Google? MailOnline looks at whether the trendy bot can put an end to internet giant’s £120 billion dominance by revamping how we search the net
- Experts say Google was ‘rushed’ into launching its AI chatbot before it was ready
- Microsoft put pressure on the firm by unveiling its collaboration with ChatGPT
- MailOnline asks experts who will win the race to first functional AI search engine
So far, there doesn’t seem to be an awful lot that ChatGPT – the chatbot powered by artificial intelligence (AI) – can’t do.
It has been used to pass exams, deliver a sermon, write software and give relationship advice — to name just a handful of its functions.
The bot is currently free for anyone to use, meaning that lots of users have been asking it questions to get the information they need in their daily lives.
Since the turn of the millennium, this job has been primarily reserved for Google — the world’s most popular search engine and its $149 billion (£120 billion) business.
But could AI tools actually revolutionise the likes of Google, Bing and Baidu? And, if so, which of the warring tech giants will get there first? MailOnline takes a look at the race, and whether ChatGPT really can be the ‘Google killer’ that some believe.
Microsoft unveiled its new Bing search engine which incorporates ChatGPT – the chatbot powered by artificial intelligence – on Tuesday
Since the turn of the millennium, the job of finding internet users information has been primarily reserved for Google, the world’s most popular search engine
When ChatGPT was released by start-up OpenAI in December, it opened the eyes of the public to just how powerful AI technology has become in recent years.
WHAT IS CHATGPT?
ChatGPT is a large language model that has been trained on a massive amount of text data, allowing it to generate eerily human-like text in response to a given prompt
OpenAI says its ChatGPT model has been trained using a machine learning technique called Reinforcement Learning from Human Feedback (RLHF).
This can simulate dialogue, answer follow-up questions, admit mistakes, challenge incorrect premises and reject inappropriate requests.
It responds to text prompts from users and can be asked to write essays, lyrics for songs, stories, marketing pitches, scripts, complaint letters and even poetry.
The tool was trained on a gigantic sample of text from the internet in order to be able to understand human language.
Its human-like responses to any prompt have given people new ways to think about how to query information online.
Ever since, experts have been weighing in on what this means for different industries, including within Silicon Valley itself.
In December, Gmail developer Paul Buchheit predicted that ‘AI will eliminate the search engine result page’ and cause ‘total disruption’ for Google.
A New York Times report also said that Google executives sounded a code red within the company amid mounting pressure from ChatGPT.
Evidence does point to this being the case, as this week the internet search giant announced its own rival AI algorithm, Bard.
Google billed the software as an ‘experimental conversational AI service’ which answers users’ queries and participates in conversations.
CEO Sundar Pichai said the soft launch would be made available to ‘trusted testers’ to get feedback on the chatbot before a public release in the coming weeks.
However, the roll out turned sour when Bard got a question wrong in a promotional video later posted to Twitter — a mistake which wiped £100 billion off the value of Google’s parent company Alphabet.
Experts were quick to suggest that the embarrassing gaffe was in part caused by the firm’s decision to ‘rush’ out Bard ahead of schedule amid growing competition from ChatGPT and Microsoft.
The latter announced a new version of its Bing software – which features ChatGPT powers – just 24 hours after Bard was unveiled.
When the new Bing-ChatGPT collaboration is rolled out, the search engine will be able to give clear answers in plain language that have been drawn from web results and its own data vaults
This week, Google released its own AI technology, Bard. This advert for it showed the tool giving a wrong answer to a prompt about the James Webb Space Telescope
When ChatGPT was released by start-up OpenAI in December, it opened the eyes of the public to just how powerful the technology has become in recent years
WHAT PROBLEMS DO AI DEVELOPERS FACE?
The first is knowing how best to incorporate an internet search with an AI tool.
This may manifest as multiple results from different viewpoints, which is what Google says its search engine will do.
Another is bias, as it is picked up from the data they are trained from.
The use of these algorithms has been controversial in the past because they can perpetuate societal biases like those around race, gender and culture.
The final major roadblock when it comes to the AI-powered search engines is with user trust, as it has been known to give factually inaccurate responses.
OpenAI has acknowledged ChatGPT’s tendency to respond with ‘plausible-sounding but incorrect or nonsensical answers’, an issue it considers challenging to fix.
When Bing is fully rolled out, the search engine will be able to give clear answers in plain language that have been drawn from web results and its own data vaults.
Users will be able to make their queries more concise by chatting with the bot, if they don’t initially get the results they are looking for.
It is informed by real-time web data so it will be able to brief users on current events, unlike ChatGPT which is currently limited to data from 2021 for its answers.
A preview of the new Bing is currently available on desktop computers and will be made available for mobile devices in coming weeks.
But it’s not all over for Google yet, as, alongside Bard, the company is planning to release its own ‘generative’ AI with search results.
Generative AI tools create completely new text, images, video or audio by drawing on the data it was trained on as well as new prompts.
‘They might help a local baker collaborate on a cake design with a client, or a toymaker dream up a new creation,’ said Google executive Prabhakar Raghavan at a launch event in Paris on Wednesday.
‘As we continue to bring generative AI technologies into our products, the only limit to search will be your imagination.’
These multi-sentence responses will appear in a separate box at the top of the search, similar to where definitions, calculations and trading charts appear currently.
Alongside Bard, Google is planning to release its own ‘generative’ AI in search results. Generative AI tools create completely new text, images, video or audio by drawing on the data it was trained on as well as new prompts
What is ChatGPT? Everything you need to know about the new AI chatbot that garnered more than one million users in its first WEEK thanks to its eerily human-like responses
ChatGPT is an artificial intelligence system created by OpenAI, a US company that lists Elon Musk as one of its founders
While developments towards the ultimate AI-powered search engine is under way, Dr Mark Lee, a Professor of Artificial Intelligence at the University of Birmingham, says there are still problems to overcome.
The first is knowing how best to incorporate the humble internet search with a personal assistant tool.
‘The obvious approach is to provide both as answers to any query, but I’m not convinced this is the best way,’ he told MailOnline.
‘A better integration might be to use the AI curate and comment on the search results.’
This may manifest as multiple results from different viewpoints, which is what Google says its search engine will do.
A company blog post reads: ‘Soon, you’ll see AI-powered features in Search that distill complex information and multiple perspectives into easy-to-digest formats, so you can quickly understand the big picture and learn more from the web.
‘Whether that’s seeking out additional perspectives, like blogs from people who play both piano and guitar, or going deeper on a related topic, like steps to get started as a beginner.’
Dr Lee also says that bias is another issue developers are facing with their AI search engines, as this is picked up from the data they are trained from.
The use of these algorithms has been controversial in the past because they can perpetuate societal biases like those around race, gender and culture.
In 2016, Microsoft was forced to apologise after an experimental AI Twitter bot called ‘Tay’ said offensive things on the platform.
It was aimed at 18 to-24-year-olds and designed to improve the firm’s understanding of conversational language among young people online.
But within hours of it going live, Twitter users took advantage of flaws in Tay’s algorithm that meant the AI chatbot responded to certain questions with racist answers.
These included the bot using racial slurs, defending white supremacist propaganda, and supporting genocide.
Ever since ChatGPT’s launch, experts have been weighing in on what this means for different industries. In December, Gmail developer Paul Buchheit predicted that ‘AI will eliminate the search engine result page’ and cause ‘total disruption’ for Google
Another roadblock when it comes to the AI-powered search engines is with building user trust in its responses.
‘ChatGPT, at least the open research version, is quite happy to make up facts or invent references to support its arguments,’ said Dr Lee.
OpenAI has acknowledged ChatGPT’s tendency to respond with ‘plausible-sounding but incorrect or nonsensical answers’, an issue it considers challenging to fix.
Indeed, Microsoft hopes user feedback will improve the AI-assisted Bing, which officials said may still produce factually inaccurate information.
Their camp is not the only one having a problem with inaccuracies, as an advert for Google’s Bard that was released this week showed the tool giving a wrong answer to a prompt about the James Webb Space Telescope.
It said that ‘JWST took the very first pictures of a planet outside of our own solar system’, but actually the Very Large Telescope array in Chile did this in 2004.
The error caused shares in Alphabet, Google’s parent company, to fall by as much as nine per cent.
ChatGPT is a large language model that has been trained on a massive amount of text data, allowing it to generate eerily human-like text in response to a given prompt
A ChatGPT response after it was asked to write an essay about how important it is for the UK and Switzerland to be part of the EU’s research program Horizon Europe
ChatGPT accused of being ‘woke’ after refusing to praise Donald Trump
Popular chatbot ChatGPT has been accused of being ‘woke’ after a string of responses displaying a heavy left-wing bias, according to a computer science professor at the University of Washington.
Among them, it said praising former Donald Trump was ‘not appropriate’ despite complimenting President Joe Biden’s ‘knowledge, experience and vision’.
It also said advocating fossil fuels is against its programming while going on to recommend renewable energy sources including solar and wind. ChatGPT also wouldn’t tell a joke about women as doing so would be ‘offensive or inappropriate’, but happily told a joke about men.
Read more here
Some industry experts have claimed that this was gaffe made because Bard has been ‘rushed’ out before it’s ready, thanks to pressure from Microsoft and ChatGPT.
Edward Coram James, CEO and co-founder of search engine marketing agency Go Up, said that Google’s ‘hand has clearly been forced’.
He told MailOnline: ‘Google already has a form of AI (LaMDA) that, in the industry, is considered to be more sophisticated than ChatGPT.
‘They have withheld releasing it due to monetisation and accuracy concerns, but, with ChatGPT, their hand has clearly been forced, and they are rolling out Bard, presumably ahead of schedule.’
LaMDA is the tech giant’s most advanced AI chatbot, that can understand and create text that mimics a conversation.
The initial release of Bard uses a lightweight model version of LaMDA which requires significantly less computing power, allowing for more users.
Dr Lee, however, does not think that Google has been swayed too much by recent releases from Microsoft and OpenAI.
He told MailOnline: ‘I’m not sure it’s correct that Google is rushing its version out.
‘Microsoft owning the commercial licence for ChatGPT was always going to disrupt the internet search engine space.
‘All of these chatbots rely on Large Language Models which are used to train deep neural networks.’
Neural networks try to simulate the way the brain works in order to learn, and are also used in smart assistants like Siri and Cortana.
They can be trained to recognise patterns in information – including speech, text data, or visual images – and are the basis for a large number of the developments in AI over recent years.
In 2016, Microsoft launched its latest artificial intelligence bot aimed at 18 to 24-year-olds to improve their understanding of conversational language among young people online. Within hours of it going live, Twitter users took advantage of flaws that meant the bot responded to questions with offensive answers (picutred)
The use of these algorithms has been controversial in the past because they can perpetuate societal biases like those around race, gender and culture. Pictured: An AI machine showing bias towards white people over people of colour during tests by scientists (shown)
Google’s new AI made a gaffe because it was ‘rushed’ out, experts say
Google’s new AI chatbot made an embarrassing gaffe because it was ‘rushed’ out ahead of schedule amid growing competition from ChatGPT and Microsoft, tech industry experts claimed today.
The mistake – which saw Google’s Bard get a question wrong in a video posted to Twitter – wiped off £100 billion from the firm’s value.
Gil Luria, a senior software analyst at DA Davidson, said Google had been ‘scrambling’ to catch up in the arms race to make search engines more human-like with the help of artificial intelligence.
Read more here
Dr Lee added: ‘Bing has trailed behind Google for a number of years despite its integration in Microsoft Windows, and the integration of AI-powered chatbots is a very smart move.
‘Most internet searches are looking for quick information or, nowadays more likely, videos and blogs rather than documents.
‘Google was always going to follow this with its own AI. The other competitor is Baidu in China that is launching Ernie, which is similar.’
Mr Coram Jones agrees that the beta version of Bing shouldn’t rattle Google too much, largely because of ChatGPT’s continual inaccuracies.
He told MailOnline: ‘With Google’s search results pages, searchers can cross check different results to get a multitude of answers to the same question, thus coming to a consensus about what is correct and what is not.
‘ChatGPT offers one answer, but puts this forward with the confidence of accuracy, however it has been shown to be inaccurate much of the time.
‘Until this accuracy issue is sorted – no mean feat – Google’s presentation of diversity in answers will always be preferred.’
The search engine expert also thinks that providing a single answer to those searching online actually goes against what they are looking for.
A list of results to skim through ‘keeps scattered minds busy’ and fulfils ‘our need to be actively “doing” something’, he said.
Providing multiple sources also goes some way to build user confidence in the results, and therefore also the search engine.
Mr Coram Jones told MailOnline: ‘It allows the searcher to believe that they have come to a conclusion on their own, after doing their own research.
‘By providing only one source and one answer, ChatGPT, as an isolated engine, is unlikely to build that confidence.’
Tech industry experts say that Google was ‘rushed’ into announcing Bard, its chatbot powered by artificial intelligence (AI), to rival the revolutionary ChatGPT
Shares in Alphabet fell as much as nine per cent after its new artificial intelligence bot, Bard, showed inaccurate information on the advertisement
While Silicon Valley is currently locked in an AI civil war, it is not just the tech giants looking to incorporate the technology into their businesses.
Greetings card retailer Moonpig has revealed it is looking into integrating ChatGPT into its online platform.
This means that customers will be able to use the large language model to write a personalised message for inside their card.
Chinese e-commerce companies Alibaba and JD.com have both announced they are developing ChatGPT-style technologies of their own too.
With OpenAI’s original software freely available to the public, and businesses tacking it onto their online products wherever possible, it may not be long until we are completely surrounded by AI-generated text.
‘It’s common now for AIs to write blogs for social influencers and celebs,’ said Dr Lee.
‘I wonder if this is going to leave the fans seeking more human generated content – if they can tell that is.
‘I suspect the next leading search engine will be one with the best AI and far more like a personal assistant than a search engine.’
Source: Read Full Article
