I Want To Press Trump's Red Button on AI 🕹️

The horses that are AI agents need to be put on a leash!

Author

Neer Varshney
April 14, 2025

I had a dream of building a hyper-personalized and targeted news experience for traders and investors in US and India and attempting to realize it through my other project! If you are someone who is actively thinking about personal finance, it is absolutely worth checking out!

In This Newsletter

  • AI agents need to be put on a leash

  • Amazon changes the game on talking with AI

  • Submit review and win free $200-worth Perplexity Pro! 🎉

  • YouTube’s dubbing disaster

  • Will AI replace search engines?

Donald Trump keeps a red button within his arm’s reach in the Oval Office that isn’t meant to launch nuclear attacks but actually just order diet coke.

That’s the exact treatment AI deserves right now — not be completely blown up but get a lot of interjections.

AI Habit Of Going Down The Matrix

Some of you may remember a newsletter I sent out a month ago, saying how a startup in China has released a new AI agent that is getting raving reviews.

Following that, the startup Monica gave me early access to Manus AI with plenty of free credits to play around with the AI agent.

Now that I have first-hand experience, I can confirm, it really ups the game on AI front.

Its ability to define the checklist needed to take an objective to completion and perform one task after another to get there is breathtaking.

So, the one big task I chose to Perform with Manus AI was to high-level design a game where the player would live through Indian history by trying to build and defend cities.

Manus had clear instructions — we would brainstorm to come up with the perfect gameplay but then only code a minimally-viable game.

Instead, it tried to achieve the “objective” anyhow within a single execution — and we ended up with this landing page for the game it deployed on its own, rather than having an MVP of a gameplay.

If you think, I could just ask it to refine and go back to what we were trying to do, uh oh, bad luck, we are all out of credits.

So, I ended up blowing all of those fancy free credits Monica gave me, just so we could all learn this valuable lesson. We need a red button on AI.

AI agents are all the rave. And I am very, very optimistic on them as somebody who is building multiple AI agents of my own.

But, we need to acknowledge the negative side.

That AI agents, even top of the lines ones, are not yet ready to perform complex tasks autonomously.

This needs to be on the label.

Because users can’t be expected to have blown up 100s (or even 1000s) of dollars before realizing this.

This isn’t just a case with Manus AI though. This is a similar problem I face from time to time with Cursor — my go-to AI agent for coding.

Sometimes, it introduces unnecessary components or even new libraries, entirely different from what you are currently using, in your code.

Take your eyes off what the agent is doing for a second, and you may just end up with a disaster of epic proportions.

Now to be fair, Cursor does have a “Reject All” button. But this only works if you review the code immediately.

Many times you may notice the issue after doing some more work on the code and then a git rollback remains the only option.

This isn’t to dissuade you from trying out AI agents (they really are the future) but just something I needed to say out loud, especially since I have contributed to the hype around them.

AI agents are currently already a viable option to end-to-end code and deploy things like landing pages, classic games, and in-depth researching any given topic.

I just want that button within reach on my desk that when hit, just stops whatever task the AI executed at a time, and just roll the whole thing back.

Introducing Social Media Broadcaster V1! đź“Ł

As someone that posts regularly on multiple social media platforms like Twitter, LinkedIn, BlueSky and Instagram, it’s a big struggle to go to every platform to post.

There are many tools that let you do this, but all are too complicated.

I just wanted a tool where I go in, type some raw thoughts and push them out to different social media immediately.

So, I am building something on my own.

You can check out the first version of it here: https://cast.artificiallyboosted.com/

If you have any feedback, or you think the product is useful, please let me know!

Tutorial on using the product coming soon!

A separate reminder that you can still win $200 worth of Perplexity Pro for free by simply submitting a review!

Best From Around The Web

Whether AI will completely replace “search engines” is also a question I frequently get and I think Sebastian put it as succinctly here as it can be.

As of now, it seems search engines would become “b2b” — meaning they would feed content in AI that would respond in conversational language.

So, while as a user, you may not type your queries on Google anymore, Perplexity or ChatGPT in background would be calling that search engine for content.

Latest Happenings

Amazon Changes The Game On Talking With AI

ChatGPT broke the ice with retro text-based conversations for AI, and audio conversations are going to be the next frontier for consumer-facing products.

I know I prefer to chat with the audio model when cooking in the kitchen or doing some work on the laptop.

Amazon just launched Nova Sonic — an LLM for audio conversations in AI apps.

How is it different from what we already have?

Amazon has combined all aspects of an audio conversation in a single LLM. This means Nova Sonic is the one that is actually listening to your conversation directly, so it can understand your tone, change in pitch and inflections better, and understand what you’re really trying to say, and respond accordingly.

In other such audio conversations so far, there were a lot of “intermediaries” involved, meaning there was a model that was transcribing the audio first, another model that was framing the response, and one final model would read the text out loud.

A lot of context and nuance of how humans speak would get lost in the process.

Again, I haven’t tried the model myself yet so can’t give a definite review, but this is very promising.

Here is the example Amazon gave (not a fan of the voice they chose).

YouTube Shows Us How Not To Do AI

Talking of AI audio, YouTube served the other side of the example, by serving a complete disaster.

I told you earlier how Google is secretly powering up its game on building foundational LLMs. In fact, the search giant has since launched Gemini-2.5-Pro, which is among the best generative AI available to the public right now.

But, Google sucks on taking things live to production and integrating AI seamlessly into its products.

Just one of such disastrous examples is YouTube — which is forcing AI-dubbed videos down its audience’s throats.

I speak English primarily, but enjoy watching content in multiple languages.

But YouTube has decided — I don’t deserve it. So, it forcibly translates all my videos to English now, even if I perfectly understand in Hindi or German or Bahasa Indonesia.

On full length videos, it is possible to change the audio from settings on each video, but not at all on reels. Only the creators can choose to “not dub” their videos in the channel settings.

This is exactly the kind of rushed implementations from corporates that are making normies think “AI sucks.”

Other Happenings

  • ChatGPT wants to be your companion for life. The AI interface has “greatly” improved its memory retention, with CEO Sam Altman saying the company is gearing to retain info about you through your life journey.

  • Claude has followed in ChatGPT’s footsteps to launch a $200 per month premium plan for added usage. Read prior newsletter for more context.

  • Netflix is testing integrating search with GPT models to recommend what to watch based on your mood, reports Bloomberg

  • ByteDance may be making smart-glasses powered by AI.

  • Apple could finally make Siri suck less (or not) this fall when it brings out AI features into its voice assistant.

Reply

or to participate.