Steven Johnson is a very meta author. He writes frequently about science and technology, and likes to immerse himself in the things he’s covering, even using them to change the way he writes books. A couple of years ago, a few months before ChatGPT launched and the AI boom took over the tech world, Johnson got a magazine assignment that sent him really, really deep down the AI rabbit hole. And he never came back up.

Now, in addition to writing books, Johnson is also working at Google. He’s part of the team building a product called NotebookLM — “Notebook,” as the team calls it. It’s a note-taking and research tool: you upload documents and import web links, and Notebook’s Gemini-powered AI helps you organize things, extract information, and understand a subject better. “They reached out,” Johnson says when I ask how he got involved with Google, “and said, ‘hey, you’ve been dreaming of this ideal software tool that helps you organize your thoughts and helps you write and helps you formulate connections and brainstorm. We think we can do it now.” Johnson signed up, and has been at Google since the summer of 2022.

The product itself first launched in 2023 as Project Tailwind, and has since been rebranded and expanded in big ways. Just last week, the team launched Audio Overviews, which generates a podcast — with two chatty hosts, plenty of back and forth, and a truly remarkable penchant for the phrases “deep dive” and “buckle up” — based on the information you provide. It’s fascinating, it’s complicated, and it’s getting better really fast.

On this episode of The Vergecast, Johnson joins to discuss his fascination with AI, his time at Google, and the present and future of NotebookLM. We talk about the complicated issues raised by a tool like this, and whether it’s okay to let an AI do your research and homework.

We also talk about how to make sure a tool like NotebookLM is both accurate and easily fact-checked, why context windows are more important to the future of AI than most people realize, and how often AI podcast hosts should say “like” in conversation. And we talk about Johnson’s own process as a writer and creator, and how AI is changing the way he works.

If you want to know more on everything we discuss in this episode, here are some links to get you started:

Phishing attacks remain a significant threat to organizations across various industries, with threat actors disguising malicious emails as legitimate messages from trusted brands, in an attempt to deceive users into divulging sensitive information or installing malware.

A multi-year survey by Cofense has highlighted phishing campaigns that spoofed trusted brands across the top 10 industries they serve – with Microsoft coming out on top as the most impersonated brand.

Did you notice that the image above was created by artificial intelligence? It can be difficult to spot AI-generated images, video, audio and text at a time when technological advances are making them increasingly indistinguishable from much human-created content, leaving us open to manipulation by disinformation. But by knowing the current state of the AI technologies used to create misinformation, and the range of telltale signs that what you are looking at might be fake, you can help protect yourself from being taken in.

World leaders are concerned. According to a report by the World Economic Forum, misinformation and disinformation may “radically disrupt electoral processes in several economies over the next two years”, while easier access to AI tools “have already enabled an explosion in falsified information and so-called ‘synthetic’ content, from sophisticated voice cloning to counterfeit websites”.

The terms misinformation and disinformation both refer to false or inaccurate information, but disinformation is that which is deliberately intended to deceive or mislead.

“The issue with AI-powered disinformation is the scale, speed and ease with which campaigns can be launched,” says Hany Farid at the University of California, Berkeley. “These attacks will no longer take state-sponsored actors or well-financed organisations – a single individual with access to some modest computing power can create massive amounts of fake content.”

He says that generative AI (see glossary, below) is “polluting the entire information ecosystem, casting everything we read, see and hear into doubt”. He says his research suggests that, in many cases, AI-generated images and audio are “nearly indistinguishable from reality”.

However, research by Farid and others reveals that there are strategies you can follow to reduce your risk of falling for social media misinformation or disinformation created by AI.

How to spot fake AI images

Remember seeing a photo of Pope Francis wearing a puffer jacket? Such fake AI images have become more common as new tools based on diffusion models (see glossary, below) have allowed anyone to start churning out images from simple text prompts. One study by Nicholas Dufour at Google and his colleagues found a rapid increase in the proportion of AI-generated images in fact-checked misinformation claims from early 2023 onwards.

“Nowadays, media literacy requires AI literacy,” says Negar Kamali at Northwestern University in Illinois. In a 2024 study, she and her colleagues identified five different categories of errors in AI-generated images (outlined below) and provided guidance on how people can spot these for themselves. The good news is that their research suggests people are currently about 70 per cent accurate at detecting fake AI images of people. You can use their online image test to assess your own sleuthing skills.

5 common types of errors in AI-generated images:

1. Sociocultural implausibilities: Is the scene depicting rare, unusual or surprising behaviour for certain cultures or historical figures?
2. Anatomical implausibilities: Take a close look: are body parts like hands unusually shaped or sized? Do the eyes or mouths look strange? Have any body parts merged?
3. Stylistic artefacts: Does the image look unnatural, almost too perfect or stylistic? Does the background look odd or like it is missing something? Is the lighting strange or variable?
4. Functional implausibilities: Do any objects look bizarre or like they might not be real or work? For example, are buttons or belt buckles in weird places?
5. Violations of physics: Are shadows pointing in different directions? Are mirror reflections consistent with the world depicted within the image?

How to identify video deepfakes

AI technology known as generative adversarial networks (see glossary, below) has allowed tech-savvy individuals to create video deepfakes since 2014 – digitally manipulating existing videos of people to swap in different faces, create new facial expressions and insert new spoken audio aligned with matching lip-syncing. This has enabled a growing array of scammers, state-backed hackers and internet users to produce video deepfakes where celebrities such as Taylor Swift and ordinary people alike may find themselves unwillingly featured in non-consensual deepfake pornography, scams and political misinformation or disinformation.

The techniques for spotting AI fake images (see above) can be applied to suspect videos too. Additionally, researchers at the Massachusetts Institute of Technology and Northwestern University in Illinois have compiled some tips for how to spot such deepfakes, but they have acknowledged that there is no fool-proof method that always works.

6 tips for spotting AI-generated video:

1. Mouth and lip movements: Are there moments when the video and audio aren’t completely synced?
2. Anatomical glitches: Does the face or body look weird or move unnaturally?
3. Face: Look for inconsistencies in face smoothness or wrinkles around the forehead and cheeks, along with facial moles.
4. Lighting: Is the lighting inconsistent? Do shadows behave as you would expect? Pay particular attention to a person’s eyes, eyebrows and glasses.
5. Hair: Does facial hair look weird or move in strange ways?
6. Blinking: Too much or too little blinking could be a sign of a deepfake.

A newer category of video deepfakes is based on diffusion models (see glossary, below) – the same AI technology behind many image generators – that can create completely AI-generated video clips based on text prompts. Companies are already testing and releasing commercial versions of AI video generators that could make it easy for anyone to do this without needing special technical knowledge. So far, the resulting videos tend to feature distorted faces or bizarre body movements.

“These AI-generated videos are probably easier for people to detect than images, because there is a lot of movement and there is a lot more opportunity for AI-generated artefacts and impossibilities,” says Kamali.

How to identify AI bots

Social media accounts controlled by computer bots have become common on many social media and messaging platforms. A growing number of these bots have also been taking advantage of generative AI technologies such as large language models (see glossary, below) since 2022. These make it both easy and cheap to churn out AI-written content through thousands of bots that is grammatically correct and convincingly customised to different situations.

It has become much easier “to customise these large language models for specific audiences with specific messages”, says Paul Brenner at the University of Notre Dame in Indiana.

Brenner and his colleagues have found in their research that volunteers could only distinguish AI-powered bots from humans about 42 per cent of the time – despite the participants being told they were potentially interacting with bots. You can test your own bot detection skills here.

Some strategies can help identify less sophisticated AI bots, says Brenner.

5 ways to determine whether a social media account is an AI bot:

1. Emojis and hashtags: Excessive use of these can be a sign.
2. Uncommon phrasing, word choices or analogies: Unusual wording could indicate an AI bot.
3. Repetition and structure: Bots may use repeated wording that follows similar or rigid forms and they may overuse certain slang terms.
4. Ask questions: These can reveal a bot’s lack of knowledge about a topic – particularly when it comes to local places and situations.
5. Assume the worst: If a social media account isn’t a personal contact and their identity hasn’t been clearly validated or verified, it could well be an AI bot.

How to detect audio cloning and speech deepfakes

Voice cloning (see glossary, below) AI tools have made it easy to generate new spoken audio that can mimic practically anyone. This has led to the rise of audio deepfake scams that clone the voices of family members, company executives and political leaders such as US President Joe Biden. These can be much more difficult to identify compared with AI-generated videos or images.

“Voice cloning is particularly challenging to distinguish between real and fake because there aren’t visual components to support our brains in making that decision,” says Rachel Tobac, co-founder of SocialProof Security, a white-hat hacking organisation.

Detecting such AI audio deepfakes can be especially tricky when they are used in video and phone calls. But there are some common-sense steps you can follow to distinguish authentic humans from AI-generated voices.

4 steps for recognising if audio has been cloned or faked using AI:

1. Public figures: If the audio clip is of an elected official or celebrity, check if what they are saying is consistent with what has already been publicly reported or shared about their views and behaviour.
2. Look for inconsistencies: Compare the audio clip with previously authenticated video or audio clips that feature the same person’s voice. Are there any inconsistencies in the sound of their voice or their speech mannerisms?
3. Awkward silences: If you are listening to a phone call or voicemail and the speaker is taking unusually long pauses while speaking, they may be using AI-powered voice cloning technology.
4. Weird and wordy: Any robotic speech patterns or an unusually verbose manner of speaking could indicate that someone is using a combination of voice cloning to mimic a person’s voice and a large language model to generate the exact wording.

The technology will only get better

As it stands, there are no consistent rules that can always distinguish AI-generated content from authentic human content. AI models capable of generating text, images, video and audio will almost certainly continue to improve and they can often quickly produce authentic-seeming content without any obvious artefacts or mistakes. “Be politely paranoid and realise that AI has been manipulating and fabricating pictures, videos and audio fast – we’re talking completed in 30 seconds or less,” says Tobac. “This makes it easy for malicious individuals who are looking to trick folks to turn around AI-generated disinformation quickly, hitting social media within minutes of breaking news.”

While it is important to hone your eye for AI-generated false information and learn to ask more questions of what you read, see and hear, ultimately this won’t be enough to stop harm and the responsibility to detect fakes can’t fall fully on individuals. Farid is among researchers who say that government regulators must hold to account the largest tech companies – along with start-ups backed by prominent Silicon Valley investors – that have developed many of the tools that are flooding the internet with fake AI-generated content. “Technology is not neutral,” says Farid. “This line that the technology sector has sold us that somehow they don’t have to absorb liability where every other industry does, I simply reject it.”

Google just announced the third generation of Wear OS-powered devices since its return to the smartwatch domain. We say devices because this time around, there are two variants — a standard 41mm model and a larger 45mm version.

Apart from the additional bigger size option, the Pixel Watch 3 looks nearly identical to last year’s Pixel Watch 2. So, if you bought the latter in 2023, does upgrading make sense? Or, if you’ve been eyeing the Pixel Watch, you could get a discounted price on the previous generation, so should you buy the Pixel Watch 2 instead?

These are two very important questions. Let us answer them for you in the comparison below. 

Google Pixel Watch 3 vs. Pixel Watch 2: specs

Google Pixel Watch 3 vs. Pixel Watch 2: design and display

The Google Pixel Watch 3 comes in a circular outer body without many visible changes from last year’s model. The addition of a bigger 45mm is perhaps the biggest change in hardware we see this year. Despite the wider landscape, the 45mm is as thick as the smaller 41mm variant, which remains largely unchanged from last year. We found that the underside of the Pixel Watch 3 does pick up a lot of scratches very easily, so it’s worth keeping this in mind.

The watch has a rotating crown/button combo placed almost flush with the body, which is made of 100% recycled aluminum, and comes in clear silver, champagne gold, or matte black finishes paired to the same band colors last year — except for a new “Rose Quartz” that is limited to the smaller 41mm variant. The Bay Blue band option will no longer be an option in the default set but you can purchase other blue bands — besides chains, metal links, fabrics, and a whole range of options — separately.

Advertisement

While the design lacks any radical changes over the previous generation, the display gets a major upgrade. Both the 41mm and the 45mm models get a newer variant of Google’s “Actua” displays with a peak brightness of 2,000 nits. That’s twice as bright as the 1,000-nit display on the Pixel Watch 2. In addition to getting brighter, the display can also get as low as 1 nit on the always-on display (AOD), so it doesn’t scorch your eyes in the dark. 

The new display also gets variable refresh rate that reaches as low as 1Hz when needed. The primary objective here is to reduce battery consumption, especially with an AOD. 

Advertisement

In addition, the displays feature slimmer bezels. Google says the 41mm’s display is now 10% bigger than last year despite any changes to the overall dimensions. Irrespective of their sizes, both displays get the same level of sharpness at 320 ppi (pixels per inch), so you shouldn’t have any trouble reading text, though interacting with the watch, and especially typing on a full-sized keyboard, is likely to be better on the 45mm variant. The bigger variant’s screen is 40% larger than that of the previous models, and this makes for a much better viewing experience, including seeing more information in notifications.

Google Pixel Watch 3 vs. Pixel Watch 2: hardware

Advertisement

Other than the display, there aren’t any noticeable changes to the hardware on the Pixel Watch 3. Google doesn’t explicitly list any upgrades to the set of sensors on the Pixel Watch 3 over the last generation, and we can take this silence as a sign of no reasonable upgrades. 

That said, the Pixel Watch 2 already had a versatile set of sensors. In addition to the standard heart rate and blood-oxygen monitor sensors, the Pixel Watch 2 — and now the Watch 3 — features sensors to monitor skin temperature and ECG. Meanwhile, another dedicated sensor to measure electrical conductance in the skin can detect changes in mood and alert you in moments of stress. Though being reminded of your worries may not be useful for everyone, it may be helpful for people who suffer physiological conditions with stress as a sign of worsening symptoms.

Advertisement

Inside, the Pixel Watch 3 still uses a Snapdragon Wear 5100 chip with an M33 co-processor for low-level tasks such as AOD. The two-chip configuration helps limit the load on the main chip, thereby reducing the battery consumption. It’s backed up by a 2GB RAM and 32GB of storage, the same as the previous generation.  

Google Pixel Watch 3 vs. Pixel Watch 2: fitness tracking and software

Although the Pixel Watch 3 lacks any notable changes to sensors for fitness tracking, Google advertises a wide range of improvements using AI. Google claims the improved training algorithms, part of the Fitbit Premium experience, now offer more precise recommendations and insight so you can adjust the intensity of your training. The Pixel Watch 3 will also give you a rundown of your condition at the start of the day with Fitbit’s Morning Brief functionality. This includes insights about your sleep quality, any overnight changes to your vitals, and the weather. 

Advertisement

Additionally, you will now be able to create routines with the Pixel Watch 3, including a sequence of difference workouts and buffers in between for rest or cooldown. You can also desired targets for distance, heart rate, and similar metrics, and the Pixel Watch 3 ensures you keep up or slow down accordingly by using audio-based and haptic reminders. For running-related activities, the Pixel Watch 3 is gaining additional metrics such as vertical oscillation and duration of contact with the ground.  

Furthermore, the Pixel Watch 3 now shows a Readiness score with an improved AI while Cardio Load tells you the ideal amount of activity you should pursue on any specific day based on various metrics, including recovery from sleep and similar workouts in the past. 

Advertisement

There are also new running recommendations powered by AI that use your Readiness Score, Target Load and running preference to provide a daily recommended run plan. As we found in our Pixel Watch 3 review, having your watch tell you precisely the kind of run to undertake is incredibly helpful in improving the variety of your running, making warm-ups more consistent, and ensuring you can enjoy running more. Like detailed sleep tracking, mindfulness sessions, guided workouts and recipes, this feature is locked behind Fitbit Premium — which costs $10 per month or $80 per year — although you will get six months for free when you buy the Pixel Watch 3.

The Pixel Watch 3 runs the newest version of Wear OS 5, although you may not notice many visual differences from the older Wear OS 4. In addition to the already existing Google Home controls on the Pixel Watch 2, the Watch 3 also gain controls to your TVs and dongles running Google TV, which presumably also includes the new Streamer box Google announced last week. In addition, you can also view and control your Nest cameras, including being able to talk to people through a Nest doorbell.

Since all of these features are available through software, Google might roll them out to the Pixel Watch 2 or even the older first-gen Watch when they receive the Wear OS 5.0 update. 

Google Pixel Watch 3 vs. Pixel Watch 2: battery and charging

Advertisement

The smaller 41mm Pixel Watch 3 gets the same 309mAh battery as the older Pixel Watch 2, which can last over a day of usage with the AOD enabled. The larger 45mm variant comes with a bigger 420mAh battery adjusted for the higher power requirements of the larger display. For both of these sizes, Google claims a 24-hour battery life with an AOD and 36 hours with the battery saver. 

In our testing, we’ve found that the Pixel Watch 3 regularly exceeds its stated claims, a surprise given this rarely happens. The Pixel Watch 3 lasts around a day and a half between charges with the always-on-display enabled, multiple tracked workouts, sleep tracking overnight, and notifications being delivered all day. This is a definite improvement over the Pixel Watch 2.

Advertisement

The Pixel Watch 3 comes the same magnetic puck with physical pogo pin connectors for charging. Without disclosing the exact charging speeds, Google says charging is 20% faster than last year. As per official claims, the 41mm variant charges 80% in 35 minutes (compared to 43 minutes earlier) while a full charge takes an hour. The larger variant will take slightly longer, owing to the larger battery. Google says the bigger Pixel Watch 3 can charge 80% in 50 minutes, while a full charge takes 80 minutes. 

Once again, we will test these claims in real-life scenarios during our review. 

Advertisement

Google Pixel Watch 3 vs. Pixel Watch 2: price and availability

The Pixel Watch 3 is available now. The smaller 41mm Pixel Watch 3 is priced at $349 without cellular connectivity and $449 with it. The larger 45mm variant is available for $399 and $449 without and with LTE, respectively. 

In comparison, the Pixel Watch 2 launched originally for $350 and $400 for Wi-Fi only and cellular variants. However, over the course of the months since its launch, the prices have dipped, and even reached as low as $270 for the sale events such as the recent Amazon Prime Day in July. 

Advertisement

Google Pixel Watch 3 vs. Pixel Watch 2: Should you upgrade?

The Pixel Watch 3 barely picks up any new features over the Watch 2, apart from a variant with a larger display. That means if you already own a Pixel Watch 2, upgrading makes little sense. 

Advertisement

However, if you want to buy one and are deciding between the Pixel Watch 3 and the Watch 2, we recommend going for the former. In addition to a new version of Wear OS and improved health metrics and AI-based recommendations, the Pixel Watch 3 will also get an additional year of software support from Google. And if you haven’t bought the Pixel Watch previously due to its smaller size, the new 45mm variant solves that too. Other reasons to choose the Pixel Watch 3 over the Watch 2 include the slightly faster battery backup and the brighter screen. 

As we covered in the review, Google has undeniably refined the experience with the Pixel Watch 3. It fixes a few lingering issues from the second-gen model while adding useful features. The result is one of the best smartwatches you can buy right now.

According to a previous report, Google AI will process unemployment claims in Nevada. In the latest news, Google CEO Sundar Pichai shares his thoughts on the possibility of AI replacing entry-level programmers. According to the CEO, it’s not the most likely scenario. Sundar Pichai has launched the 2024-25 President’s Lecture Series at Carnegie Mellon.

He traveled to Carnegie Mellon University to tour research labs, talk with faculty and students about emerging technologies, and officially inaugurate the first President’s Lecture Series event of the new academic year. The company has already taken steps towards transparency in the Gen AI era.

Google’s CEO believes even entry-level programmers can view AI as a collaborative tool

According to Google’s CEO, AI should be viewed more as a collaborative tool instead of something to compete against. Upon being asked by students whether AI would replace entry-level programmers, Pichai started out by stating that “there’s a version of this question which can be asked across many disciplines.”

He added that “the most likely scenario in all of these things is, it will help people. It’ll both help existing programmers do their jobs, where most of their energy and time is going into, you know, higher aspects of the task. Rather than you know fixing a bug over and over again or something like that, right.”

The CEO says that he has already experienced the efficiency gains at Google. Citing the AI-powered code editor Cursor AI, he said, “It is just lowering the barriers for who can program, right, like how can you, more, in a natural language medium, interact. So, programming becomes more like a creative tool. I think that’s gonna enable and make it accessible to more people.”

For those unfamiliar, Cursor AI is an AI-powered code editor that helps developers write better code more efficiently. It offers a variety of features to help with coding, including intelligent code suggestions, code generation, and bug identification. There is a free Hobby tier, a Pro tier for $20 per month, and a Business tier for $40 per user per month.

Google is also prioritizing the use of clean energy for AI development

He not didn’t only talk about his views on the use of AI but also about the role of clean energy in its development. It involves powering massive data centers. He acknowledged the growing interest in small modular reactors (SMRs) for nuclear energy. Capital investment was reportedly flowing into developing alternate energy sources to cleanly power computing facilities.

He joined Google in 2004 and helped lead the development of Google Toolbar. Later he led the development of Google Chrome, which would become the world’s most popular internet browser. Back in August 2015, he became the CEO of Google, and in December 2019, the CEO of Alphabet.

Rumors emerged last year of a collaboration between former Apple designer Jony Ive and OpenAI CEO Sam Altman, but the two have until now kept quiet about it. In a profile by that was published this weekend, though, Ive confirms his company LoveFrom is leading the design on an AI product being built with Altman. Also on board are Tang Tan and Evans Hankey, both of whom held big design roles at Apple.

There’s so far a team of about 10 employees involved with the project, based in a San Francisco office building that’s one of several properties Ive has purchased on a single city block, according to the Times. But we still don’t know much about the product they’re working on. The report describes Tan and Hankey wheeling chairs between the LoveFrom properties that were “topped by papers and cardboard boxes with the earliest ideas for a product that uses A.I. to create a computing experience that is less socially disruptive than the iPhone.”

Since Ive left Apple in 2019 to start LoveFrom, the design firm has worked on and even , but we haven’t seen much in the way of hardware — just . While an AI product seems to be on the horizon, there’s currently no timeline for when it’ll make its debut.