Duolingo goes AI-first, but faces backlash from millions of users

Welcome to this edition of Loop!

After taking a few months to travel around Australia and Asia, I’ve been slowly getting back into the swing of things.

Thank you to everyone for your support over the last year. With thousands of people now reading my insights every week - from business leaders, to engineers, to creatives, and everyone in-between - I really do appreciate your support.

Soon, I will be launching Loop Academy - which is a collection of in-depth guides around emerging technology and how you can apply them in your work. Keep your eyes peeled for that email in the coming days.

To kick off your week, I’ve rounded-up the most important technology and AI updates that you should know about.

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

• Why Duolingo is facing a huge backlash from their users and what business leaders can learn from it

• IBM’s plans to spend a whopping $150 billion in the US

• Anthropic’s huge upgrade to Claude that lets you connect your own data

• … and much more

Let's jump in!

1. Duolingo faces a backlash from users, after saying it will replace workers with AI

Earlier in the week, Duolingo’s CEO announced that the company is shifting to an "AI-first" approach.

As a result, Duolingo will use AI to reduce its reliance on contractors and will only hire new staff if a task “cannot be automated”.

Luis von Ahn said that this is being done to “remove bottlenecks” and complete more work, but Duolingo has faced a strong backlash from users.

In recent years, the company has done an incredible job at marketing itself on social media - with constant memes used to gain millions of followers on TikTok.

Most of these followers are from Generation Z (born between 1997 and 2012) and have fueled the company’s growth. In the last year alone, it has seen revenues grow by a whopping 38%.

However, this is a double-edged sword. Gen Z are incredibly active online, which can both boost a company’s growth and also destroy it when they campaign against you.

Following the announcement that Duolingo will use AI to replace staff, many users have reacted angrily at the level of corporate greed - with some deciding to delete the app entirely.

Of course, it’s hard to know how widespread this is and whether it will translate into reality - but some of these videos have quickly reached over 7 million people.

It’s a lesson that companies should not be caught up in the “technology bubble”. While some investors and executives like to see these announcements, most of the general public absolutely hate it and are (understandably) fearful about what it means for their jobs.

2. Microsoft made an ad with generative AI and nobody noticed

In a sign of what’s to come, Microsoft used generative AI to create an incredibly convincing ad and released it three months ago - but nobody noticed that it wasn’t real.

Intricate movements, such as close-ups of hands typing on keyboards, were shot live, while quick cuts or shots with limited motion were created with generative AI tools.

The team used AI tools to generate the script, storyboards, and a pitch deck. They then iterated through thousands of different prompts to generate the videos, until they were happy with the results they had.

Using generative AI helped Microsoft’s team to reduce their project’s costs by over 90%, which is surprising given how much prompting was needed to create the final videos.

This shows just how rapidly the technology is advancing, and when used skillfully by production crews, can create content that looks incredibly realistic.

3. IBM plans to spend $150 billion in the US

The enterprise giant has pledged that it will invest over $150 billion in the United States, spread across the next five years.

More than $30 billion of this will be spent on mainframe and quantum computing research, as IBM aims to solidify its position in the sector.

It’s worth noting that IBM are not alone, as other technology firms are also planning to invest huge sums in the US - such as Apple ($500 billion), OpenAI ($500 billion), Nvidia ($500 billion), SoftBank ($100 billion), Microsoft ($80 billion), and Google ($75 billion).

Most of those companies are investing in data centres for upcoming AI systems, but IBM’s focus on quantum computing is an interesting change.

Given that Quantinuum has made several breakthroughs with quantum, it seems like a good time for IBM to invest more in the technology.

4. Driverless trucks are now making deliveries in Texas

Aurora is now running its autonomous trucks between Dallas and Houston, with over 1,200 miles already completed without a driver.

By the end of 2025, the company will expand its driverless service to El Paso and Phoenix as well.

It was once thought that driverless trucks would be operating before robotaxis, but hurdles around technology and regulation have led to delays.

Aurora has spent four years conducting supervised pilot hauls, mostly in Texas, delivering over 10,000 customer loads across 3 million autonomous miles.

It’s a promising startup that’s worth watching. They’ve recently signed partnerships with leading companies in the trucking industry - including Continental, Volvo, and Uber.

But given the safety issues that Cruise has faced with their autonomous vehicles, Aurora will have to tread extremely carefully if they want to avoid a similar fate.

5. Amazon says it can weather the new US tariffs

In Amazon's Q1 2025 earnings call, CEO Andy Jassy was confident that the company can handle the Trump administration's worldwide tariffs.

As of right now, the company hasn’t seen a significant decrease in demand from consumers - or an increase in retail prices from sellers - but that might change in the next few weeks.

It’s worth stressing that the tariffs pose challenges for most US retailers, but Amazon believes that its huge size and global presence will help to keep prices low.

Given that they made $59 billion in profit last year, the company could absorb some of the tariff charges to stabilise prices.

Interestingly, Apple also sounded similarly confident. Most of the devices that they sell in the US will be manufactured in either India or Vietnam - which are facing much lower tariff rates than China is.

For the rest of the world, many of their devices will still be built in China - but obviously won’t face those 145% tariffs since they’re not bound for the US.

Nvidia’s new tool can turn 3D scenes into AI images

The company has unveiled a new tool, which allows developers to generate AI images from a 3D scene.

To demonstrate that Nvidia is just as bad as OpenAI at naming their products, they called it “AI Blueprint for 3D-guided generative AI”. Yes, really.

Despite the weird name, it’s a really fascinating tool. The main issue we have with image generators is that we lack control over the final output.

All we can do is write a text prompt and hope for the best. It’s not a great user experience.

To tackle this, Nvidia’s blueprint tool allows you to view your scene in 3D and adjust where objects are placed within it - which offers us a lot more control over the finer details.

They were able to achieve this by combining Blender's 3D modelling software with the popular FLUX image generator.

If you want to try out the tool yourself, I’ve included a link to it below.

Anthropic lets you connect apps to Claude

In the last few days, Anthropic has announced a huge upgrade for their Claude chatbot and will allow you to easily connect apps to the AI model.

Strangely, this seems to have slipped under the radar and isn’t getting as much attention as it deserves.

For example, you can now connect Claude to your Jira / Confluence data and ask the model questions - rather than having to copy and paste the content yourself.

Or small business owners could connect their Square payments system to the model, then ask it to analyse how well their business has performed this week.

This is all possible thanks to Anthropic’s Model Context Protocol (MCP), which you might have heard about before.

Essentially, it makes it incredibly easy for us to connect these models to our data and has standardised the entire process.

As more companies start to embrace the standard, we will have even more options to choose from and speed up the way we work.

Today, you can get started with lots of different data sources - such as Atlassian’s Jira and Confluence, Zapier, Cloudflare, Intercom, Asana, Square, Sentry, PayPal, Linear, and Plaid.

It’s expected that the new “integrations” feature will roll out for most paid users soon, with those on Max, Team, and Enterprise plans already given access.

🍎 Apple is now moving US iPhone production to India

🤖 Manus, the Chinese AI startup, gets new funding at a $500 million valuation

🖥️ Nvidia will invest $500 billion to develop AI supercomputers in US

🛒 ChatGPT is now better at shopping, while Visa and Mastercard unveil their own AI tools

🚕 Waymo is considering a partnership with Toyota on autonomous vehicles

🚀 Amazon launches Nova Premier, its most capable AI model yet

👋 Intel's chief commercial officer resigns, as the company’s future looks bleak

🗣️ Google’s new AI tools allow you to learn languages in a more personalised way

📹 NYC wants AI subway cameras to predict “trouble” before it happens

👓 Meta changes privacy policy for Ray-Ban glasses, so it can train AI models

💰 Jeff Bezos plans to sell billions in Amazon stock

🧠 Microsoft's tiny AI model, Phi 4, can rival the performance of far larger systems

Mach Industries

This is an impressive defence tech startup that was founded by a 21-year-old, soon after he dropped out of MIT.

Mach Industries is close to securing $100 million in new funding from investors, which will value the company at around $470 million.

The startup is based in Huntington Beach, California and is focused on developing its own vertical takeoff (VTOL) aircraft.

In March, they were selected by the US’ Army Applications Laboratory - which is the US military’s innovation unit - to develop a vertical takeoff precision cruise missile called "Strategic Strike".

Mach is also planning to build its first factory in California, which will cover a huge 115,000-square-foot to develop these weapons.

Overall, the company has raised an impressive $185 million to date. They will be competing with other defence tech startups, such as Anduril and Shield AI, when trying to secure government contracts.

We’re seeing a clear trend here, with investors much more willing to fund defence startups. This is largely due to the impact of the Ukraine-Russia war in Europe, which has accelerated the use of advanced drones and new weapons systems on the battlefield.

This Week’s Art

We’ve covered quite a bit this week, including:

• The backlash that Duolingo faced for going “AI-first”

• Microsoft’s convincing ad that was made with generative AI

• IBM’s plans to spend $150 billion in the US

• Aurora’s operation of driving trucks in Texas

• Why Amazon is optimistic about President Trump’s tariffs

• Nvidia’s new tool that can turn 3D scenes into AI images

• How you can connect dozens of apps to Claude

• And how Mach Industries is trying to compete with Anduril for US military contracts

Thank you again to everyone for your support. If you found something interesting in this week’s edition, please feel free to share this newsletter with your colleagues.

Have a good week!

Liam

Share with Others

If you found something interesting in this week’s edition, feel free to share this newsletter with your colleagues.

About the Author

Liam McCormick is a Senior AI Engineer and works within Kainos' Innovation team. He identifies business value in emerging technologies, implements them, and then shares these insights with others.