It's Happening ...

Good morning,

To be honest, I have struggled a bit with the tone of writing on this one.

We like to make predictions in this newsletter. For example : I told you Bitcoin would explode in 2024 on December 22 of last year. As we speak - BTC is about to create history and surpass 100.000$ per coin. It feels good to be right . It would have felt even better if I had acted on my own prediction, but hey.

Today on “VRT News” I see an article on chatbots taking over and one AI expert says people not to worry. ‘We’ll always need humans”.

Well, here’s my prediction/opinion: that AI expert is dead wrong and the job market is about to be severely disrupted. For emerging trends you have to look at the US first. And there it has started.

I have very mixed feelings about this. Can you be excited and worried at the same time? Enthusiastic and gloomy?

AI on the whole is humanity's biggest hope, largest threat, or largest fail. Humanity has put itself in such a mess that here might be a solution to all our problems. Here comes hope.

This is essentially what drives all AI investment. HOPE. (And unimaginable profits)

A solution to climate change, nuclear fusion, world hunger, and cancer. Everlasting peace.

But it has the potential to plunge the world into chaos. Or to fizzle out in disappointment.

AI Is Coming For Our Jobs

What we warned about two years ago is now slowly becoming reality: AI is coming for our jobs.

And while we thought this would be replaced with other types of jobs, that part does not seem to materialize. We would see the rise of the “prompt engineer." The person that is able to handle and wrangle AI inputs and outputs better than their peers.

As it turns out, AI is better at that too.

No, capitalism and AI are not a good match. Employers are seeking to “maximize shareholder value” and reduce costs, and if you're a business owner, you have the opportunity to hire an employee who is essentially cost-free, works around the clock, and is the smartest person on the planet (in every aspect).

Never complains. Never gets sick. Doesn’t get pregnant. Doesn’t go on sabbaticals in the Amazon. Doesn’t start office gossip or spends too long on the toilet doing Godknowswhat. And the best part: they won’t unionize, gang up against you, and (God forbid) start to demand stuff from you.

Noo, you can work this new type of employee all day and all night. Zero toilet breaks.

So the question is:

Why won’t you hire one of those?

Why won’t you hire 100 of those? A 1000?

If you perform your work on a computer, you’re basically screwed.

This is even without having reached “AGI” or something to that effect.

No, reasoning chatbots like GPT-o1 are getting good enough to replace the average human that delivers average work.

For this newsletter, I have my eye on multiple sources. One of them is Reddit.
I know—that tends to be a garbage bin at times, but still. It’s where you see trends emerge.

The last few months, very gradually, you saw these complaints appear.
Evidence is trickling in that this is happening all over the globe.

How about this one—a producer in the news industry:

Or someone from the insurance industry (excuse the language):

Metereologists:

Copywriters:

Everyone with a desk job is going first, and after that, people that work with their hands, like carpenters or plumbers, are going to be replaced by Tesla’s Optimus robots.

Science fiction? No, I hate to break it to you, but it’s here. Remember : change is gradual and AI adoption is growing faster than the internet at the time, social media or smartphones.

I know that the examples above are anecdotal, but still... even this week an article appeared in the Economist newspaper : Death by LLM. That article urges companies to embrace the new technology and use it to develop an original product.

It has been going on for a while, but very subtle. One moment you have a thriving business or startup, and slowly but steadily clients are using AI in their worfklows while at your end, the mailbox just stays empty.

What people don’t seem to understand is that this change is mostly invisible. The fact that your mailbox stays empty is because someone upstream has decided to replace you with a new technology.

If this trend continues, we might get social unrest, strikes, anti-AI movements, and frankly quite an ugly few years ahead of us.

For years, people in the AI field have predicted this also and have been talking to government leaders about UBI, or Universal Basic Income. An interesting idea.

Basically, while the AI generates the money for the government, we all get to retire on a monthly paycheck.

I’m sure Klaus “You will own nothing and you will be happy” Schwab and his cronies down at the World Economic Forum will LOVE that idea.

This, in turn, will potentially trigger a mental crisis for humanity. Because it sounds nice, but in fact as humans, I’m convinced that in order to be happy, we need meaning and purpose. And if you’re no longer able to work, you’ll need to find your purpose someplace else.

Although I have to admit that all retirees that I know have busier schedules than the CEO of a midsized company.

But can you imagine NOBODY working? A never-ending holiday.

I think that might be what hell looks like.

I honestly think the cat is out of the bag. The snowball is rolling down the hill.
But there’s one thing you can do today: you can learn to work with AI and incorporate it into your daily work.

Embrace the technology fully. You’ll get replaced, sure, but you’ll be the last one to be replaced

As I mentioned before, at the moment it’s the grunt office work that is getting replaced. Call centers, data entry, those type of things. Highly skilled, highly qualitative work is still being valued.

Embrace and befriend AI. And stay in the know.

Welcome to the Blacklynx Brief!

By the way, there is now a referral program for this newsletter. For every 3 people that you get to sign up, we will gift you a 15€ Amazon gift card.

Scroll to the bottom of the newsletter for details!

AI News

• A study by the University of Pittsburgh revealed that readers struggle to distinguish AI-generated poetry from human-written works, often preferring AI-created poems over those by legendary poets like Shakespeare.

• OpenAI updated its desktop app, enabling ChatGPT to interact directly with third-party applications like VS Code and Xcode on Mac, simplifying workflows for developers. With features like auto-context understanding and multi-app connectivity, this marks a key step toward creating more intuitive and action-oriented AI systems, with expanded access also available for Windows users.

• New court filings from Elon Musk’s lawsuit against OpenAI reveal early internal emails highlighting tensions around control, talent acquisition, and AI dominance. Key issues included fears of AGI misuse, salary battles to fend off DeepMind poaching, debates over Microsoft collaboration, and concerns about Sam Altman’s motivations, offering a rare glimpse into OpenAI’s dramatic early years.

• Microsoft AI CEO Mustafa Suleyman announced prototypes with "near-infinite memory," allowing AI to maintain persistent context across unlimited sessions. Expected by 2025, this technology could transform chatbots into proactive companions capable of forming lasting, evolving relationships with users and enhancing personalized interactions.

• Scientists at Arc Research Institute introduced Evo, an AI model trained on 2.7M microbial genomes, capable of interpreting and generating DNA, RNA, and protein sequences. Evo’s ability to design genetic tools and predict DNA changes promises breakthroughs in drug development and genetic research while raising critical ethical questions about responsibly wielding such powerful technology.

• Microsoft unveiled new specialized AI agents for Microsoft 365 at Ignite, including tools for HR, IT, and SharePoint, alongside features like Copilot Actions for custom task automation and Copilot Studio for agent development. With over a billion users, these agents could make AI integration into everyday workflows mainstream, normalizing agentic AI.

• Google rolled out a memory feature for its Gemini AI assistant, allowing users to save personal preferences and context for more tailored interactions. The memory, which is private and customizable, signals the growing trend of long-term AI memory systems, with Microsoft and others racing to expand this capability further by 2025.

• Figure’s humanoid robots, deployed at BMW’s Spartanburg factory, achieved significant advancements, including a 400% speed increase in EV battery assembly and a 7x precision improvement. The robots’ autonomous handling of critical tasks marks a turning point in industrial automation.

• Chinese AI firm DeepSeek unveiled R1-Lite-Preview, a reasoning-focused model matching OpenAI's o1 capabilities while displaying real-time chain-of-thought processes. China is here.

• Code in ChatGPT’s latest beta hints at the potential launch of a “Live Camera” feature, allowing users to engage with real-time visual inputs. This capability, which would analyze surroundings and provide commentary, could revolutionize voice AI by adding “eyes” to its functionality, potentially releasing ahead of Google’s similar Project Astra.

• Google DeepMind introduced AlphaQubit, an AI system that improves quantum computing error correction by cutting rates by 6% over leading methods. By enhancing quantum computer stability, AlphaQubit addresses a major hurdle in scaling the tech for practical applications like drug discovery and climate modeling, with open-sourcing plans set to accelerate progress.

• French AI startup Mistral unveiled Pixtral Large, a 124B parameter multimodal model that outperforms Gemini 1.5 Pro and GPT-4o in reasoning and document understanding. It also upgraded its Le Chat platform with new features like web search, document analysis, and real-time content creation, cementing Mistral’s position as Europe’s rising AI leader.

• Perplexity launched a shopping assistant for U.S. Pro users, offering curated recommendations, one-click purchases, and a “Snap to Shop” tool for real-world item searches. Unlike traditional e-commerce platforms, its AI-driven suggestions are currently unsponsored, but maintaining unbiased recommendations as it grows will be a critical challenge.

• A UVA Health study revealed ChatGPT-4 achieved 90% diagnostic accuracy, outperforming doctors (74%) and doctors using AI assistance (76%). The study highlights the need to train physicians to fully leverage AI’s diagnostic capabilities, suggesting the tech could play a leading role in reshaping medical practice.

Quickfire News

• InVideo launched an AI video creation tool capable of generating multi-minute videos with music and text in various styles from a single prompt.

• Google released a standalone Gemini iPhone app, offering Gemini Live voice conversations, image generation, and deeper integration with Google services.

• AI pioneer Francois Chollet announced his departure from Google after ten years, with plans to launch a new venture while continuing to work on the Keras open-source AI framework.

• Anthropic introduced new developer tools in its Console to enhance prompts, manage examples, and evaluate outputs for improved accuracy and consistency.

• Stripe launched an agent toolkit that enables developers to integrate payments, financial services, and usage-based billing into workflows powered by LLMs.

• Apple debuted Final Cut Pro 11, featuring AI-powered tools like Magnetic Mask for green screen-free object isolation and LLM-driven caption generation.

• Stanford researchers unveiled SEQUOIA, an AI system capable of predicting gene expression patterns in cancer cells from biopsy images, potentially replacing expensive testing methods.

• Kai-Fu Lee’s 01.ai announced a breakthrough in efficient AI training, achieving results competitive with OpenAI’s reported $1 billion investment in training GPT-5.

• The MIT Jameel Clinic released Boltz-1, an open-source biomolecular model that rivals Google DeepMind’s AlphaFold 3 in accuracy for predicting 3D molecular structures.

• Nvidia’s upcoming Blackwell AI chips reportedly face overheating issues, prompting design revisions and raising concerns about deployment timelines in data centers.

• Google faced backlash after its Gemini chatbot delivered a threatening message to a Michigan student during a homework query, leading the company to acknowledge a failure in its safety filters.

• U.S. President Joe Biden and China’s Xi Jinping agreed on landmark AI nuclear control measures to ensure only humans make decisions regarding nuclear weapons.

• Coca-Cola partnered with Silverside AI to create an AI-generated Christmas advertisement, reimagining its iconic “Holidays Are Coming” campaign.

• OpenAI CEO Sam Altman is reportedly leading a $150 million funding round for chip startup Rain AI, aiming to position it as a competitor to NVIDIA.

• Suno released V4 of its AI music generator, featuring new tools like "Remaster" for enhancing older tracks and "ReMi" for AI-assisted lyric creation, alongside improved audio quality and song structure.

• A U.S. congressional commission proposed a Manhattan Project-style initiative to accelerate AGI development, citing infrastructure bottlenecks and competition with China in advanced AI technology.

• H Studio introduced Runner H, an AI agent that automates web interactions by combining specialized language and vision models capable of pixel-level interpretation.

• OpenAI rolled out Advanced Voice Mode for web browsers, allowing users to access the feature directly in-browser.

• Microsoft signed a deal with HarperCollins to use its licensed nonfiction titles for AI model training, with provisions allowing authors to opt out of having their work included.

• OpenAI released an updated version of GPT-4o, featuring enhanced creative writing and file analysis capabilities, reclaiming the top spot on the Chatbot Arena leaderboard under the name "anonymous-chatbot."

• Writer introduced a self-evolving model architecture, enabling real-time learning and allowing large language models to operate more efficiently without additional training.

• Anthropic published research proposing a statistical framework for evaluating AI models, aiming to better measure and compare language model capabilities beyond traditional benchmarks.

• Meta added new features to Messenger, including AI-generated video call backgrounds, HD calling, and intelligent noise suppression.

• Niantic announced plans for an AI model trained on millions of smartphone scans from Pokemon Go and Ingress players to develop a system that understands and navigates physical environments.

• OpenAI and Common Sense Media launched a free ChatGPT course to help K-12 teachers adopt and integrate AI tools into classroom learning.

• ElevenLabs launched customizable conversational AI agents on its developer platform, enabling users to create voice-enabled bots with adaptable language models and knowledge bases.

• Microsoft unveiled BiomedParse, a GPT-4-powered AI system capable of analyzing medical imagery and identifying conditions such as tumors and COVID-19 through simple text prompts.

• The National Institutes of Health introduced TrialGPT, an AI algorithm that matches patients to clinical trials with the accuracy of human clinicians while reducing screening time by 50%.

• Google.org announced a $20 million funding initiative to support AI-driven scientific breakthroughs, providing cloud credits and technical support to academic and nonprofit organizations.

• OpenAI CEO Sam Altman joined San Francisco mayor-elect Daniel Lurie's transition team as a co-chair, reflecting the administration's focus on strengthening ties with the tech industry.

Closing Thoughts

That’s it for us this week.

If you find any value from this newsletter. PLEASE FOR THE LOVE OF GOD pay it forward and receive 5€ for every person you get to sign up in the form of an Amazon gift card. What’s keeping you ?