політика, бізнес, народ

Google announced Tuesday that its Bard chatbot would be integrated into Gmail, YouTube and other applications in a push to broaden Alphabet’s user experience.

Google has spent years refining its generative AI without immediate plans to release a chatbot, until OpenAI unveiled ChatGPT late last year and partnered with Microsoft to popularize the cutting-edge tool. Google scrambled to put together its response: Bard.

Google cleared hurdles earlier this year to release Bard across the globe in dozens of languages, squeaking past European regulators who raised questions about the chatbot’s effect on data security.

The search engine giant is now waging a campaign to win public support.

These new updates — Bard extensions — represent the company’s most ambitious attempt at popularizing generative AI. Going forward, Bard can work as a plug-in with Google Drive, Gmail, YouTube and more.

A user might ask Bard to distill a string of lengthy and confusing emails into a pithy summary or order the chatbot to find the quickest route to an address using Google Maps.

The plug-in can be used by students and professionals who might want Bard to scour dense PDFs and Google Docs and return a list of bullet points.

A common criticism of chatbots is their inaccuracy and apparent ability to falsify information. Computer scientists call this flaw “hallucinations.” The Bard plug-in will include a button to fact-check the chatbot’s answers against search engine results in real time to determine if Bard is “hallucinating.”

Generative AI combs vast databases for linguistic patterns and other information in a process known as data-scraping. Data-scraping is what empowers Bard and ChatGPT to create unique, humanlike answers to queries in an instant. Essentially, chatbots imitate what is already available on the internet.

Activists have long worried that companies might train their chatbots on unsuspecting users’ personal information. Google said that Bard will access private data only with permission.

Google also said that any data-scraping it might perform on what users have stored in their personal Docs, Drive or Gmail accounts would not be used in targeted advertising or training Bard. Nor would private content be accessible to Google employees.

“You’re always in control of your privacy settings when deciding how you want to use these extensions, and you can turn them off at any time,” Google said in a blog post.

The Bard extensions come after Microsoft similarly incorporated ChatGPT into Bing earlier this year but ultimately failed to gain ground in its war on Google’s search engine dominance.

According to market analytics, ChatGPT, Bard’s top competitor, has been suffering marked declines in its user base as mania over generative AI has waned in recent months. Google is hoping to capitalize on ChatGPT’s losses and for Bard to catch up.

Some information for this story came from Agence France-Presse.
…

Britain has invited China to its global artificial intelligence summit in November, with foreign minister James Cleverly saying the risks of the technology could not be contained if one of its leading players was absent.

“We cannot keep the UK public safe from the risks of AI if we exclude one of the leading nations in AI tech,” Cleverly said in a statement on Tuesday.  

Prime Minister Rishi Sunak wants Britain to become a global leader in AI regulation and the summit on Nov. 1-2 will bring together governments, tech companies and academics to discuss the risks posed by the powerful new technology.

Britain said the event would touch on topics such as how AI could undermine biosecurity as well as how the technology could be used for public good, for example in safer transport.  

Cleverly, who last month became the most senior minister to visit China in five years, has argued for deeper engagement with Beijing, saying it would be a mistake to try to isolate the world’s second largest economy and Chinese help was needed in areas such as climate change and economic instability.

“The UK’s approach to China is to protect our institutions and infrastructure, align with partners and engage where it is in the UK’s national interest,” Cleverly said on Tuesday.  

London is trying to improve ties with Beijing but there has been growing anxiety about Chinese activity in Britain in recent weeks after it was revealed that a parliamentary researcher was arrested in March on suspicion of spying for China.

The Chinese embassy in London was not immediately able to say if China would attend the AI summit.

Britain has appointed tech expert Matt Clifford and former senior diplomat Jonathan Black to lead preparations for the summit.  

The Financial Times reported that government officials want a less “draconian” approach to regulating the technology, compared with the European Union’s wide-sweeping AI Act.  

Under the incoming EU legislation, organizations using AI systems deemed “high risk” will be expected to complete rigorous risk assessments, log their activities, and make sensitive internal data available to authorities upon request.  

Clifford told Reuters last month that he hoped the UK summit would set the tone for future international debates on AI regulation.
…

Just as many in the United States are starting to explore how to use artificial intelligence to make their lives easier, U.S. adversaries and criminal gangs are moving forward with plans to exploit the technology at Americans’ expense.

FBI Director Christopher Wray issued the warning Monday, telling a cybersecurity conference in Washington that artificial intelligence, or AI, “is ripe for potential abuses.”

“Criminals and hostile foreign governments are already exploiting that technology,” Wray said, without sharing specifics.

“While generative AI can certainly save law-abiding citizens time by automating tasks, it can also make it easier for bad guys to do things like generate deepfakes and malicious code and can provide a tool for threat actors to develop increasingly powerful, sophisticated, customizable and scalable capabilities,” he said.

Wray said the FBI is working to identify and track those using AI to harm U.S. citizens but added that the bureau is being cautious about employing AI itself.

“To stay ahead of the threat at the FBI, we’re determining how we can ethically and legally leverage AI to do our jobs,” he said.

When contacted by VOA, the FBI declined to elaborate on its concerns about employing AI. Nor did the bureau say when or if it has used AI, even on a limited basis.

Other U.S. national security agencies, however, are currently making use of AI.

The Department of Homeland Security is using AI to combat fentanyl trafficking, counter child sexual exploitation and protect critical infrastructure, according to department officials, even as they roll out guidelines governing its use.

“Artificial intelligence is a powerful tool,” Homeland Security Secretary Alejandro Mayorkas said in a statement last Thursday. “Our department must continue to keep pace with this rapidly evolving technology, and do so in a way that is transparent and respectful of the privacy, civil rights, and civil liberties of everyone we serve.”

DHS has also issued directives aimed at preventing its use of AI from being skewed by biased learning models and databases, and to give U.S. citizens a choice of opting out of systems using facial recognition technology.

But across multiple U.S. departments and agencies, the fear of the potential damage AI could cause is growing.

FBI officials, for example, warned in July that violent extremists and terrorists have been experimenting with AI to more easily build explosives.

And they said a growing number of criminals appear to be gravitating to the technology to carry out everything from petty crimes to financial heists.

It is China, though, that is driving the bulk of the concern.

National Security Agency officials have warned that Beijing started using AI to disseminate propaganda via what they described as a fake news channel last year.

“This is just the tip of the iceberg,” David Frederick, the NSA’s assistant deputy director for China, told a cybersecurity summit earlier this month.

“[Artificial intelligence] will enable more effective malign influence operations,” he added.

Such concerns have been bolstered by private cybersecurity companies.

Microsoft, for example, warned earlier this month that Chinese-linked cyber actors have started using AI to produce “eye-catching content” for disinformation efforts that has been gaining traction with U.S. voters.

“We can expect China to continue to hone this technology over time, though it remains to be seen how and when it will deploy it at scale,” Microsoft said.

For its part, China has repeatedly denied allegations it is using AI improperly.

“In recent years, some western media and think tanks have accused China of using artificial intelligence to create fake social media accounts to spread so-called ‘pro-China’ information,” Chinese Embassy spokesperson Liu Pengyu told VOA in an email, following the publication of the Microsoft report.

“Such remarks are full of prejudice and malicious speculation against China, which China firmly opposes,” Liu added.
…