Google's new AI research assistant can plan your next holiday for you – and now it's free
If you haven't heard of Deep Research, you're not alone. In Google 's race to expand its AI products across smartphones, search results and Google Workspace, the tool was easily overlooked – not least because it has until now been exclusive to the paid-tier of the Gemini app.
Launched in December 2024, Deep Research effectively searches the web on your behalf. Give it a topic you want to learn more about – from child-friendly ski resorts to nuclear reactor legislation – and the AI assistant will disappear for five minutes, trawling the internet for trusted sources and up-to-date information before popping back up with a comprehensive, in-depth report.
With today's upgrade to Deep Research 2.0 – and by throwing it open to everyone to try – Google is hoping the tool will appeal not just to academic and scientific users 'geeking out on nuclear fusion', but to the average person, whether they're planning a holiday, moving home or comparing local landscaping services.
Aarush Selvan, senior product manager at Google, gave The Independent a first look at the new version of Deep Research ahead of its relaunch. 'We're looking at the kinds of tasks where before you might have had to open 50 Chrome tabs and weave together information to get to an answer, or to go really deep on a topic,' says Selvan.
The new version of Deep Research upgrades the tool to Google's latest 'thinking' AI model – which transparently explains its reasoning as it works, and questions and corrects itself as it goes, which Google says should lead to more accurate results and a greater insight into how it arrives at its conclusions.
'This new model is actively thinking between taking steps,' says Selvan. 'That allows it to be more thoughtful in how it browses the web and to do more analysis on its findings. The idea is it will start being able to write more insightful and more detailed reports.'
To demonstrate the newly upgraded Deep Research tool, Selvan tasked the AI with looking for potential holiday destinations in the Canary Islands along with a bunch of specific, personal requirements: two adults and one seven year old, a focus on good food and entertainment, with destinations ranked by average daytime temperatures.
Whereas previously Deep Research would only display a list of the websites it was rifling through for information, the new thinking model gives a running commentary on its thought process: first identifying a longlist of possible holiday destinations, then diving into hotel websites to find kids' clubs, searching local restaurants and venues, and pulling up historical weather data to whittle down its report to a handful of relevant results.
'[Deep Research 2.0] is now able to articulate what it's found, why it's interesting, and what it should look for next. We can bring that information back and show the user,' says Selvan.
'Another great example we worked on internally was finding the best summer camp for your kid, right? Like there are all these different small camps to consider, and you want to figure out which are half-day, which are full-day, and how they're all priced.'
Today's update to Deep Research comes as Google faces increasing pushback from online publishers, who say that its controversial AI-generated search results and summaries – similar to those created by Deep Research – reduce traffic and ad revenue. Among other changes, Deep Research 2.0 at least puts a greater emphasis on identifying its sources, giving citations and outbound links to the sites being referenced. Selvan describes the assistant's reports as a 'starting point'.
'We want to make it really easy for users to double click into each website. That's why we dedicate quite a lot of screen real estate to [links]. We want people to click into these websites, read what's interesting and understand why these websites are appearing here.'
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
The Independent
11 hours ago
- The Independent
MTG again lashes out at Trump this time over his AI plan: ‘An absolute threat to federalism'
Republican Rep. Marjorie Taylor Greene, a staunch ally of the MAGA movement, broke from President Donald Trump on Thursday to oppose his executive order on artificial intelligence, raising concerns about its impact on the environment and states' rights. Hoping to accelerate the United States's development of AI, Trump signed a series of executive orders on Wednesday, making it easier to build data centers on federal lands and incentivizing states to impose fewer regulations by threatening to withhold federal funding to AI projects. But Greene raised a red flag, saying she was concerned with the impact of massive data centers on the people and environment around, while providing little to no regulations. 'My deep concerns are that the EO demands rapid AI expansion with little to no guardrails and breaks. It also contains the threat of withholding federal funds from states who regulate AI, which is an absolute threat to federalism and why I strongly opposed the AI state moratorium originally in the BBB,' Greene wrote on X. Trump has pushed for more AI development in the U.S., hoping for companies such as Meta, Google, OpenAI, and Microsoft to develop and expand their AI technology beyond what the Chinese-based company DeepSeek has already done. But the Georgia congresswoman said the 'rushed AI expansion' should include a plan to address human and environmental impact – with particular concerns around using nearby water supplies that cross state lines. 'Competing with China does not mean become like China by threatening state rights, replacing human jobs on mass scale creating mass poverty, and creating potentially devastating effects on our environment and critical water supply,' Greene wrote. 'This needs a careful and wise approach. The AI EO takes the opposite.' Greene's statement is the latest in a string of opposing stances that she has taken against Trump, whom she has typically expressed unwavering loyalty to. Recently, Greene has also criticized the administration for not releasing the Epstein Files, a recent Trump-backed crypto bill, and the president's decision to intervene in the Iran–Israel conflict by bombing Iranian nuclear facilities. Her additional opposition also comes at a moment of contention between the president and his MAGA base. Trump's dismissal of the Epstein Files, of which conspiracy theories have floated in the MAGA world for years, appears to have caused a fracture in his base's trust. It's unclear how many may feel about the AI bill, but Greene's opposition provides some insight. 'I represent the base and when I'm frustrated and upset over the direction of things, you better be clear, the base is not happy,' Greene said on X back in May.
Reuters
14 hours ago
- Reuters
Managing reputation in the age of synthetic content
July 18, 2025 - The introduction and increased usage of generative artificial intelligence (AI) have been influential throughout the legal industry, offering unprecedented opportunities for efficiency while introducing significant risks to law firm brands. Tools such as ChatGPT, Grok, Gemini, and proprietary large language models (LLMs) can pop out client memos, legal alerts, and thought leadership articles at record speed. However, as law firms embrace these technologies, they face an urgent imperative: using the power of generative AI in a way that does not compromise the integrity of their brand. Generative AI offers law firms a range of amazing efficiencies such as the ability to produce first drafts of documents, analyze vast amounts of data, and generate commentary on case law in minutes. These functions save time and are extremely valuable, particularly in fast-moving regulatory and litigation environments. As an example, AI tools can quickly draft breaking news client alerts on recently introduced legislation and summarize complex case law. This can free up an attorney's time for more strategic work as well as enhance a firm's responsiveness, positioning it as a leader in a competitive market. Yet, speed could come at a cost. AI lacks an inherent understanding of legal nuance and most importantly, ethical obligations, or jurisdictional differences that a human understands. It can operate without accountability, increasing the risk of producing inaccurate, overly generalized, or fabricated content. A notable example is a syndicated article published in the Chicago Sun-Times and Philadelphia Inquirer in summer 2025, which included a reading list featuring nonexistent books in the summer 2025 reading list article. As reported in The Washington Post, ("Major newspapers ran a summer reading list. AI made up book titles." May 20, 2025) the author admitted to using AI tools without human editing, leading to factual errors and reputational damage for both the media outlets and the author. For law firms, similar missteps could erode client trust and tarnish a brand built over decades. The Wall Street Journal certainly could have been thinking of brand awareness when they issued their own recommendations of 14 books to read. The adoption of generative AI introduces several risks that can undermine a law firm's reputation: (1) Accuracy and misinformation: AI-generated content is prone to "hallucinations," which are confidently stated but factually incorrect information. In the legal context, publishing flawed analyses, outdated citations, or incorrect interpretations of case law could have disastrous consequences, particularly if the firm's name is attached. (2) Plagiarism and intellectual property infringement: Many AI models are trained on datasets scraped from the internet. This raises the risk that outputs may inadvertently replicate existing work without proper attribution, which could expose firms to copyright infringement claims or ethical violations, both of which carry significant reputational and legal consequences. (3) Dilution of expertise and thought leadership: Overreliance on AI for client alerts, bylined articles, or white papers risks diluting a firm's unique voice and eroding its authority. We see too many headlines using the word navigating, do we not? Clients hire lawyers for their cultivated expertise, unique insight, and judgment, not for generic, machine-generated summaries that lack depth or context. (4) Reputational fallout from internal use: Even internally circulated AI-generated content, such as draft memos or research notes, can create liabilities if leaked or misused. A poorly written or inaccurate draft that reaches the public eye, whether through a data breach or accidental disclosure, can damage a firm's credibility. From a public relations perspective, generative AI requires a new framework for content governance. Law firms must move beyond enthusiasm for efficiency and embrace disciplined oversight. Here are five essential steps: •Establish a human touch policy: All AI-generated legal content should be reviewed and ideally co-authored by a licensed attorney. Firms must make it clear that no AI-generated output goes client-facing without human verification. •Implement AI disclosures and transparency protocols: Whether, in client alerts or public-facing articles, firms should consider disclosing when AI has been used in content creation. This transparency not only builds trust but protects against future claims of misrepresentation or malpractice. •Train attorneys and comms teams in AI literacy: Educate your lawyers, marketers, and PR professionals on the wonders, capabilities, and also the limitations of generative AI. Understanding how these tools work and where they can fail is critical to mitigating brand risk. •Audit for originality and attribution: Use plagiarism detection software such as Grammarly and internal checks to ensure content is original and appropriately sourced. This applies even to internal research memos and pitch materials. •Define your firm's brand voice and reinforce it: AI can write in any voice. Law firms have a tone of voice. The challenge is ensuring it consistently writes in your voice. Set clear tone, style, and messaging guidelines so that AI-generated drafts align with your firm's brand identity. Used thoughtfully, generative AI can elevate a firm's ability to communicate, respond, and engage with clients and the media. But it must never replace human insight, editorial judgment, or the firm's hard-earned reputation. Your brand is your most valuable asset, so guard it accordingly, even when the content comes from a machine. AI should serve as an enhancer, not a substitute, for the qualities that make your firm unique. Thoughtful implementation requires rigorous oversight, ethical considerations, and a steadfast commitment to authenticity.
Daily Mail
14 hours ago
- Daily Mail
Ultra-ritzy enclave where nine local families have collective wealth of $638 billion
Silicon Valley is the epicenter of America's tech hub, attracting many of the world's richest individuals and generating an estimated wealth of $638 billion. But now it has emerged that the eye-watering sum is shared between just nine families. And a new study conducted by San Jose State University concluded that the city's wealthiest households are only gaining wealth. Their collective net worth was up by a staggering $136 billion from 2024, according to the recent data. The annual report, titled the Silicon Valley Pain Index,is designed to measure 'personal and community distress' through data tracking structural inequalities in the region. Silicon Valley refers to the southern part of the San Francisco Bay Area, including Santa Clara, San Mateo, the western edge of Alameda, and Scotts Valley in Santa Cruz. It represents a high concentration of wealth and can be an increasingly difficult place for working and middle-class families to make ends meet. Meta boss Mark Zuckerberg is among the ritzy residents, accounting for $253.5 billion of the area's wealth. Zuckerberg's net worth is more than 20 times the wealth of the bottom 446,505 households in the area. The Meta CEO purchased a five-house compound in the area, estimated to be worth a whopping $37million. He bought a $7 million mansion in 2011 before buying the neighboring residences to create the compound. The report also estimates that the region has $1.01 trillion of liquid wealth. Three companies based in Silicon Valley - Adobe, Alphabet (Google), and Meta comprise a combined net worth of $535 billion. Even outside of the nine wealthiest families, the annual income needed to afford just a median-priced house is $370,000, up 54 percent in the last six years, according to the report. The report also found that renters need to have an average annual income of $136,532 to afford an apartment at 30 percent of their take-home pay, representing the highest rents in the country. In San Mateo County, a household annual income of $156,650 for a family of four is considered 'low, ' compared to the state average of $94,500. Nearly 71 percent of homes sold in the area were priced around 1.92 million and were considered the 'median home price.' The median home price in Santa Clara alone was 2.171 million in May, which is a 3.4 percent increase from last year. The study doesn't explicitly state the nine families contributing to the mass concentration of wealth in addition to Zuckerberg, but listed Google's former CEO, Larry Page, as the second richest billionaire in the area. Sergey Brin, who co-founded Google with Page and served as the president of Alphabet, is the third-wealthiest billionaire in the region. Jensen Huang, the CEO of Nvidia, an artificial intelligence company, comes in fourth. Page and Zuckerberg live in Palo Alto, while Brin and Huang have properties in Los Altos Hills. Former Google CEO Eric Schmidt is listed as the fifth-wealthiest billionaire in the region, with Dustin Moskovitz of Facebook as the sixth. Financier George Roberts was listed seventh, while Laurence Powell Jobs of Apple came in eighth. Venture capitalist John Doerr came in ninth, and Charles Schwab of Charles Schwab brokerage ranked tenth. Silicon Valley is home to major enterprises including Apple, Alphabet, Chevron, Meta, Visa, and Wells Fargo. The companies employ a significant portion of residents in the region, leading to the high concentration of wealth from executives and engineers reporting high incomes. The success of these companies has solidified Silicon Valley as a wealthy enclave and the tech capital of the world, but the region has a long history of advancement. The silicon chip, which is used in computerized machinery, was invented in the region, and electronics reporter Dan Hoefler coined the name 'Silicon Valley' in 1971.
