Skip to main content

Google plans to let software engineers use AI assistants in job interviews

Google CEO Sundar Pichai
Google CEO Sundar Pichai.
  • Google is piloting a new interview process where software engineering candidates use AI in interviews.
  • It’s part of a bigger overhaul of Google’s interview process.
  • The changes will “better align with the modern engineering landscape,” Google said in an internal doc.

Want a software job at Google? Bring your AI wingman.

The company is piloting a new interview process for software engineering candidates that will let them use an AI assistant, according to an internal document reviewed by Business Insider.

The change is part of a broader overhaul of Google’s interview process, which the document says is being made “to better align with the modern engineering landscape.”

Google will test the new format, which applies to junior to mid-level roles, to select teams in the US and plans to scale it more widely across the company and regions later if it’s successful.

Starting in the second half of the year, Google will permit the use of an “approved” AI assistant during its “code comprehension” round. Candidates will be expected to “read, debug, and optimize” an existing code database, the document states.

“Interviewers will evaluate Al fluency, including prompt engineering, output validation, and debugging skills,” it adds.

A Google spokesperson confirmed the plans and said its own AI model, Gemini, would be the AI assistant used by candidates during the pilot phase.

“We’re always evolving our interview processes to ensure we’re recruiting and hiring the best talent,” Brian Ong, vice president of recruiting at Google, told Business Insider. “As a part of that, we’re rolling out a pilot for software engineering interviews to be more reflective of how our teams are operating in the AI era.”

The new interview process reflects the significant changes that AI has brought to software developer roles. In late 2025, Anthropic and OpenAI launched new models that dramatically improved the capabilities of coding agents.

Now, three-quarters of new code created inside Google is generated by AI, the company said in April. That same shift is happening elsewhere. Greg Brockman, OpenAI’s president, recently said that AI has gone from writing 20% of code to 80%.

‘Human-led, AI-assisted’

The Google document outlines several other changes the company intends to make to its interview process, which will be piloted first.

The “Googleyness and Leadership” round — which has typically focused on behavioral questions — will now also involve a technical design discussion about a candidate’s past project.

For more junior candidates, one of the technical rounds will be replaced with an interview that will require them to “tackle open-ended engineering challenges,” the document states.

The company will pilot the new formats across several orgs, including Cloud and its platforms and devices unit, this month.

Google’s move follows what some tech startups have been embracing for a while. Graphic design giant Canva and AI coding startup Cognition are among the companies allowing candidates to use AI in technical interviews.

Emily Cohen, who heads people and operations at AI coding company Cognition, told Business Insider last week that it has changed its interview process to incorporate AI use.

“I guess this is like asking a kid to take a math test without a calculator,” she said about not allowing AI use in interviews. “For the bulk of building something similar to what you would do on the role, you can and should use AI tools.”

The document about Google’s new interview process describes it as being “human-led, AI-assisted” and says the format should better simulate a software engineer’s “workflow in the GenAl era.”

Additional reporting from Shubhangi Goel.

Have something to share? Contact this reporter via email at hlangley@businessinsider.com or Signal at 628-228-1836. Use a personal email address and a non-work device; here’s our guide to sharing information securely.

Read the original article on Business Insider

A cardiologist eats 90 grams of fiber a day without trying. Here’s her breakfast, lunch, and dinner.

danielle belardo
Dr. Danielle Belardo runs a preventive cardiology practice in Los Angeles. She said it’s easier to maintain a high-fiber diet using frozen fruits and vegetables than fresh.
  • Most Americans eat half the recommended daily amount of fiber.
  • A vegan cardiologist eats 90 grams of fiber a day thanks to her plant-based diet.
  • Fiber can lower “bad” cholesterol levels, help you feel full, and reduce the risk of cancer.

Most of us don’t eat enough fiber. Not Dr. Danielle Belardo, though.
She routinely eats a whopping 90 grams a day, more than most Americans get in a working week.

“There’s no guideline that says anyone should reach 90 grams of fiber a day, it just happens to be what I eat,” Belardo, a vegan cardiologist practicing in Los Angeles, told Business Insider.

The general recommendation is for fiber to make up 14 grams of each 1,000 calories we consume. That amounts to at least 25 grams a day or the fiber in two cups of black beans. But only 10% of Americans actually get that much fiber. Most are lucky if they get even half of that.

A high-fiber diet packed with foods such as fruits, vegetables, and whole grains, helps the body remove waste products, like bile acid, and effectively sucks “bad” cholesterol out of the bloodstream. Low-fiber diets are linked with poor heart health, higher rates of diabetes, high blood pressure, and cancers, including colon cancer and breast cancer.

But “fibermaxxing” by suddenly increasing the fiber in your diet if your body isn’t used to it can lead to bloating, gas, and stomach cramps. So, it’s advised to add fiber to your diet slowly and steadily, making sure you’re staying well-hydrated along the way.

Here’s how Dr. Belardo beats out just about everybody else when it comes to packing fiber (and protein) into her daily meals.

Breakfast: a smoothie, with berries, protein powder, and frozen spinach

blender berries for smoothie
Blackberries and raspberries are some of the highest fiber fruits.

Belardo’s typical breakfast smoothie:

  • 1 cup of frozen raspberries (8 g of fiber)
  • 1 cup of frozen blackberries (8 g)
  • 1 cup of spinach (1 g)
  • 1 scoop of protein powder (4 g)

Fiber from breakfast: 21 g

Belardo has nearly hit the daily target for women, which is 25 to 28 g, and she has only just started her day!

Lunch: avocados, beans, and whole grains

quinoa bowl with avocado and black beans
Avocados pack a fiber-rich punch, each containing about 10 grams of fiber.

At lunchtime, Belardo does some variation on a grain bowl.

A fiber-rich lunch:

  • 1 whole avocado (10 g)
  • 1/2 cup of black beans (7.5 g)
  • 1 cup of quinoa (5 g)
  • 1 cup of Brussels sprouts (4.5 g)
  • 1 Tbsp of chia seeds (5 g)

Fiber from lunch: 32 g

Total fiber consumed so far: 53 g

Dinner: pasta, chickpeas, broccoli, and beans

mung bean pasta
Belardo loves edamame and mung bean pasta, which has a greenish hue.

Bean pastas are such a great option to get high fiber in,” she said. She often recommends these to her patients looking to lose weight and improve their heart health.

Fiber fills you up, and fiber-rich foods tend to replace other, less nutrient-dense foods in a person’s diet.

“If you’re eating 30 grams of fiber from fruits and vegetables, that is naturally only going to displace something else that may be less health-promoting,” she said.

Belardo added: “You can do lentil or chickpea, but one that’s got a super high amount of protein and fiber that I love is edamame and mung bean pasta.”

A high-fiber dinner:

  • 1 1/2 cups of cooked edamame and mung bean pasta (24 g)
  • 1/2 cup of cooked chickpeas (6 g)
  • 1 cup of broccoli (5 g)
  • 1 Tbsp of black beans (1 g)

Fiber from dinner: 36 g

Total daily fiber: 89 g (Her average daily fiber intake, she said, tends to be about 90 grams, give or take).

It’s better to get fiber from foods than supplements

fiber supplement with water
Belardo says skip the fiber supplements, and amp up your fiber intake with foods like fruits, vegetables, and beans instead.

Supplements can help level out deficiencies, but Belardo doesn’t typically recommend them, because fiber-rich foods have so many benefits that simply can’t be replicated by a pill or powder.

“It’s not just the fiber itself that’s beneficial, it’s the food it comes packaged with,” Belardo said. “It’s also coming with phytonutrients, vitamins, minerals, antioxidants and all these other things.”

Read the original article on Business Insider

TikTok pulls back on an AI feature that described Charli D’Amelio as a collection of blueberries

An image of Charli D'Amelio smiling appears next to a screenshot of an AI overview of her TikTok account.
Charli D’Amelio was compared to blueberries by an AI feature on TikTok.
  • TikTok is pulling back on testing an AI feature that added text summaries to videos.
  • The tool was prone to hallucinations, comparing Charli D’Amelio to blueberries and a dog training video to origami.
  • Going forward, the feature will focus on identifying products in videos, a TikTok spokesperson said.

Did TikTok just have its “glue pizza” moment?

The company is pulling back on a new artificial intelligence feature it was testing that went haywire, adding wildly inaccurate AI-generated text summaries to videos from users like Charli D’Amelio, Shakira, and Saturday Night Live.

These “AI overviews” were designed to provide additional context for a video, recommend similar products to what’s on-screen, and generally explain what was happening. While the tool did a good job summarizing some posts, for others, it hallucinated more than Bryan Johnson did during his livestream mushroom trip.

Here are a few chaotic examples of AI overviews this reporter saw on the app in the past week:

  • The AI feature described a video of Charli D’Amelio, sitting alone in front of a white wall and talking directly to the camera, as a “collection of various blueberries with different toppings.”
  • A post from a dog trainer explaining why dogs kick their feet after going to the bathroom was described as “a captivating display of intricate origami art, meticulously folded from a single sheet.”
  • A video from Shakira promoting a new song release, per the AI overview, was “a repetitive sequence of several distinct blue shapes appearing and moving across the screen.”
  • A viral post about feeling heartbroken from a user named Victoria was described as a “mesmerizing close-up of a tiny hand repeatedly tracing intricate patterns on a smooth surface.”
  • A video of Olivia Rodrigo promoting her upcoming appearance on SNL was called a “person’s face being gradually replaced by a random, nonsensical string of letters and numbers.”

One Reddit user wrote that it was as if the feature would look at a video and then “independently open a different tab and use a random text generator to create a caption.”

Now, following user feedback, TikTok is pulling back on the test, a company spokesperson confirmed to Business Insider. The AI overview feature has been updated to focus on identifying products in a video rather than describing a video’s full contents, they said.

The tool, which has been testing for a few months, was available to a limited set of users in the US and a few other markets, the spokesperson said, describing it as an experiment.

TikTok declined to share which models it used to power its AI overviews, but an in-app description of the feature says it relied on either TikTok’s own AI tech or third-party products.

Seeing hallucinatory AI summaries show up in my feed this week felt like a throwback to when ChatGPT and other early AI products regularly made stuff up.

When Google released its version of AI overviews in 2024, the feature confidently proclaimed that a dog had played in the NHL, for example. It told my coworker Katie Notopoulos to add glue to her pizza to keep the cheese from sliding off.

While AI tools are getting better at answering some questions — a recent analysis by the AI firm Oumi found Google’s AI overviews were accurate about 90% of the time — it felt oddly comforting to see that a technology that doomsdayers predict will wipe out white-collar jobs and take over many facets of our lives can still fail in silly and surprising ways.

Read the original article on Business Insider

The US Navy shot out the rudder of an Iran blockade runner with a Super Hornet’s 20 mm cannon

An F/A-18E Super Hornet, attached to Strike Fighter Squadron (VFA) 151, launches from the flight deck of Nimitz-class aircraft carrier USS Abraham Lincoln (CVN 72) during Operation Epic Fury, March 31, 2026.
An F/A-18 Super Hornet launches from the flight deck of USS Abraham Lincoln.
  • A US Navy fighter jet disabled the rudder of an Iranian-flagged tanker on Wednesday with its 20 mm cannon.
  • The tanker was trying to bypass the blockade of Iran’s ports, the US military said.
  • It’s the second time the US has used force against an Iranian commercial ship in recent weeks.

A US Navy fighter jet knocked out the rudder of an Iranian-flagged oil tanker in the Gulf of Oman on Wednesday using its 20 mm cannon, the American military said.

The incident is the second time that the Navy has used force to disable an Iranian commercial vessel since the Trump administration began its blockade of maritime traffic going in and out of Iranian ports last month.

US Central Command, which oversees American forces in the Middle East, said it observed the unladen tanker M/T Hasna as it transited international waters en route to an Iranian port on the Gulf of Oman.

The ship “failed to comply” with repeated US military warnings that it was sailing in violation of the blockade, CENTCOM said. A Navy F/A-18 Super Hornet fighter jet that launched from the aircraft carrier USS Abraham Lincoln fired several rounds from its 20 mm cannon gun into the Hasna’s rudder.

In addition to the missiles the carrier-based fighter jet is able to carry on the wings, the Super Hornet is armed with an internal 20 mm M61A2 six-barrel rotary cannon.

CENTCOM said that it disabled the tanker, which is “no longer transiting to Iran.” It added that the blockade of Iran’s ports “remains in full effect.”

Last month, a Navy destroyer fired nine inert rounds from its 5-inch deck gun into the engine room of an Iranian-flagged cargo vessel trying to evade the blockade in the Arabian Sea. It marked a rare use of force by the Navy against a non-combat vessel.

The front of the USS Spruance is seen at night with another vessel next to it conducting an at-sea refueling.
The USS Spruance (DDG 111) is an Arleigh Burke-class guided missile destroyer.

The Navy destroyer, USS Spruance, intercepted the Iranian-flagged cargo boat M/V Touska, which was sailing toward Iran through the Arabian Sea at 17 knots. The Navy said that it issued several warnings, which the Touska ignored. The Spruance then ordered the evacuation of the engine room and opened fire.

The engagement disabled the vessel, which was subsequently boarded by Marines.

The US military operation against Iran, Operation Epic Fury, has concluded. The more recent focus has been on securing ship movement through the Strait of Hormuz as part of a new initiative called “Project Freedom,” though that was paused Wednesday as Washington and Tehran discuss a deal.

In recent days, US Navy vessels have been working to clear a path through the strait and establish a defensive umbrella for commercial ships to travel under. Iran has fired drones and missiles at commercial and military vessels in the area. The US has fended off threats and responded with force, including by using helicopters to sink Iranian small boats.

Read the original article on Business Insider

I went from no kids to twins. I’m still learning how to be a dad.

Dad with twin daughters
  • I went from zero to two kids, and you grossly underestimate experiencing life before twins.
  • There is nothing in this world to prepare you for becoming a twin parent.
  • The Married Single Parent thing is a real phenomenon, so I step up as best as I can.

In February 2019, at 12:23 and 12:24 p.m., my life changed in an instant.

On the second night after my twin girls were born, we wanted them to sleep in the room with us. That night was the exact moment when I realized that twin parenthood is a totally different beast.

We were constantly awakened because one child would fall back asleep, only for the other to wake up shortly after. That cycle continued the entire night. Needless to say, we sent them back to the nursery so we could get some sleep.

A few days later, we were discharged to go home, and the journey began: adjusting to being twin parents.

My wife and I enjoyed being DINKs

Before kids, my wife and I traveled, had plenty of late nights, bar-hopped, attended Beychella, and enjoyed the freedom of being a DINK couple. Then, in the summer of 2018, we found out we were going to be parents.

Dad holding twins
The author says having twins made him softer in a good way.

We really underestimated this and didn’t know how much life was about to change.

Nothing actually prepares you for becoming a parent. I’m seven years into this, and I’m still learning. Going from zero to two overnight was a shocker, and nobody can ever prepare you for this.

It feels like you are being herded as fast as possible through a narrow hallway that doesn’t end, and the anxiety of what happens next is always on your tail. The first, and obvious, change is finances. Everything is in twos: formula, clothes, private school tuition, everything. One of the beauties of having a gap in kids is the hand-me-down factor. With multiples, you don’t get that.

My mom did her best to teach me the ins and outs of home operations from her side of the house. Then, when parenthood came along, I went into focus mode so my wife wouldn’t feel like a married single parent.

I don’t want my wife to be a married single parent

I do my best to support my wife as much as possible. Thinking back on my childhood, I remember my dad being present. I used to hear stories about absent dads and learned to appreciate his presence, especially as I got older. I learned from my mom after he passed that he tried his best. She also made sure to instill in me what it meant to support your wife as a present parent.

Family posing for photo
The author doesn’t want his wife to be a married single parent.

Now, talking to other parents and even the organic interactions in public make being a married single parent a real problem. For example, one night while we were feeding, I noticed my wife was tired and exhausted from her day. My mother’s words and advice clicked right then. I refuse to let her be a married single parent. She’s my best friend, and I wouldn’t want to let my best friend drown.

With us both navigating twin parenthood, stepping up as best I can is an active choice every day. It’s also a choice that has taught me a lot.

My twins are everything to me

While the challenges of twin fatherhood are heavy, being a twin dad is one of the best blessings I have. My girls love coming to the hardware store with me, and our thing is playing tic-tac-toe on the smart fridges. It’s moments like that that teach me the value of just being there in the simple moments.

Another thing my daughters have taught me is how to be soft. I grew up playing sports, had a stint in combat sports, am an entrepreneur, and am always seeking a challenge. Building the plane as it’s flying is in my nature, so I developed a bit of a hard skin to withstand adversity. When they came along, I had to learn to revise that. They are full of emotions and, even now, at times, it’s a challenge for me to navigate.

My girls are helping me learn how to dial it back.

Read the original article on Business Insider

Anthropic taps Elon Musk’s SpaceX for more AI compute power

Elon Musk
Elon Musk, SpaceX CEO.
  • Anthropic and SpaceX have signed a new deal for 300 megawatts of compute capacity.
  • The OpenAI rival will use Colossus One data center to boost Claude Code for developers.
  • SpaceX, after acquiring xAI, is inking new deals to sell GPU access.

Anthropic is turning to Elon Musk for computing capacity.

As the company faces a surge in demand for its AI coding service, Claude Code, Anthropic will use SpaceX’s Colossus One data center to meet the moment. Chief Product Officer Ami Vora announced the blockbuster deal at Anthropic’s developer conference in San Francisco on Wednesday morning.

Vora said that Anthropic is doubling the current 5-hour rate limit for developers on Pro, Max, Team, and seat-based Enterprise plans. The company is also raising rate limits for developers using the advanced Claude Opus model’s API, allowing more input and output per minute. These moves will stretch the company’s already strained computing capacity.

Given this challenge, Anthropic has turned to SpaceX, which acquired Musk’s xAI lab back in February. SpaceX’s Memphis-based Colossus One data center is one of the world’s largest, housing tens of thousands of advanced Nvidia chips that run AI’s compute-intensive work. Musk’s rapid buildout of the site and the nearby Colossus 2 data center have provoked ire locally, with residents complaining of pollution from gas turbines.

Anthropic expects the deal to deliver more than 300 megawatts of computing capacity within the month, across more than 220,000 Nvidia GPUs. This builds on the company’s strategy of sourcing chip time and capacity from hyperscaling giants — work with Google, Amazon, and Microsoft is also underway.

The deal is another in SpaceX’s fledgling compute sales business. Business Insider reported in April that the coding startup Cursor would train its latest model on xAI’s GPUs. By inking computing deals with other companies, SpaceX and xAI can generate revenue from their own infrastructure, while also still developing AI models.

Musk, as recently as February, excoriated Anthropic’s Claude model as “misanthropic and evil.” But the billionaire’s mind has changed, he wrote Wednesday on X. Musk said he spent time last week with Anthropic leaders, and was impressed: “Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector.”

There could be another deal between Anthropic and SpaceX in the works. Tucked low in the press release accompanying this announcement, Anthropic wrote that it has “expressed interest” in working with SpaceX to build data centers in space.

Have a tip? Contact this reporter via email at scouncil@insider.com or Signal, Telegram, or WhatsApp at 415-757-8198. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.

Read the original article on Business Insider