AI Not Only Augments Humans - It Replaces Them
Or not
Another busy AI week went by. The race between the AI giants continues with Google’s launching Gemini 2.0, Apple integrating ChatGPT with Siri, and Microsoft releasing its new Small Language Models. And we are heading to a future, with humans with no jobs at all…
✍️ In This Issue ✍️
This Week in AI: Google's Gemini 2.0, Apple Integrates ChatGPT, OpenAI's SORA Fail, and Microsoft Offering Alternative SLM
Deep Dive: AI Will Take Your Job
AI Tools We Love: ChatGPT Screen Share🛠️
Culture Corner: Friends or Slaves?
AI Art: Optimus Awards
Quote for You: Gray Scott💡
This Week in AI
🚀 Google launched Gemini 2.0, its new AI model for practically everything. Google’s new Trillium AI chip delivers 4x speed and powers Gemini 2.0. With its new processors, Google enters the chip war against Nvidia, offering a highly affordable hardware solution for specific AI tasks, e.g. training of large models.
🌍 Midjourney is launching a multiplayer collaborative worldbuilding tool called ‘Patchwork’. I wonder when the first geeks put together the core of the 'Oasis' (have you seen the movie 'Ready Player One'? Well, you should...).
📚 Harvard and Google to release 1 million public-domain books as AI training dataset. This might easy the data scarcity issues to train new AI models.
🍎 With iOS 18.2, Apple completes its AI starter kit. ChatGPT is now integrated into IOS, and if you use Siri and she reaches her limits, she simply reaches out to ChatGPT (in case you allow it, of course).
🎥 OpenAI made SORA public. The grounbreaking text-to-video tool is now available for Plus and Pro users. Unfortunately, I can't use it in Europe, yet. Maybe in 2025. Until then, they hopefully improve it further: according to recent reviews, Sora’s AI video revolution is still a ways off.
🧠 Anthropic’s fastest model, Claude 3.5 Haiku, now generally available. The race among LLMs and AI agent platforms goes on.
Microsoft Releases New Phi-4 Model: A Small Rebel
While developers often chase ever-larger Large Language Models (LLMs) like GPT-4 (rumored to have 1.8 trillion parameters), Microsoft is betting big on Small Language Models (SLMs). These models, with billions of parameters instead of trillions, are designed for efficiency—offering lower costs and energy use while maintaining competitive performance.
What happened: Microsoft introduced Phi-4, a 14-billion-parameter model, as part of its “best quality for the cost” Phi series. Despite its smaller size, Phi-4 reportedly outperforms larger models like GPT-4o and Gemini 1.5 Pro in math benchmarks. However, it struggles with factual knowledge and complex instructions due to its size.
How it works: Microsoft’s novel approach to Phi-4 involved:
Curating high-quality data from web content, books, and code.
A refined post-training process.
Generating synthetic data using an AI pipeline, enhancing its utility for training.
The research is yet to be peer-reviewed, and details on energy use, emissions, and training data remain undisclosed. Still, Phi-4 highlights the potential of smaller models in generative AI.
Meta’s Invisible Watermarking Model: Meta Video Seal
Meta Video Seal is an AI model that enables the embedding of invisible watermarks in videos to ensure their authenticity and traceability. These watermarks are undetectable to viewers but can be later retrieved to verify the video's provenance, even after heavy editing.
How It Works
The model uses advanced algorithms to integrate watermarks into videos at a pixel level. These watermarks are resistant to typical video editing techniques like cropping, resizing, color adjustments, and even compression. Meta claims this resilience ensures that the authenticity of a video can be confirmed under various conditions.
Significance
Combatting Deepfakes: With the explosion of AI-generated deepfakes, verifying the authenticity of video content is increasingly critical. Meta Video Seal provides a robust tool to identify manipulated content, helping curb the spread of misinformation.
Content Ownership: The model can protect intellectual property by embedding ownership credentials directly into the video, preventing unauthorized use or misrepresentation.
Media Transparency: By enabling content creators to verify their work, the model promotes transparency in digital media, building trust with audiences and stakeholders.
Meta has not disclosed the energy or carbon costs of training and deploying this model, which raises concerns about its environmental impact. Additionally, the effectiveness of the technology against increasingly sophisticated manipulations remains to be tested on a large scale.
Deep Dive
AI Not Only Augments Humans - It Replaces Them
The tech layoff wave is still going strong in 2024. Following significant workforce reductions in 2022 and 2023, this year has more than 130,000 job cuts across 457 companies.
Tech layoffs in 2024
In 2024, the tech industry experienced significant upheaval, with major companies implementing widespread layoffs. Firms such as Meta, ExxonMobil, and Boeing reduced their workforces, citing economic pressures, strategic realignments, and the integration of artificial intelligence (AI) and automation as key factors.
The integration of Artificial Intelligence (AI) into various industries is transforming job markets worldwide. This wave of innovation has sparked debates about job displacement, productivity gains, and the redefinition of workplace roles. Drawing from key reports and studies, this article explores AI’s dual role as a disruptor and a creator, offering nuanced insights into its effects on labor markets.
Key Insights from Reports
1. Widespread Automation and Displacement Risks
According to the report "AI Won't Just Assist People—It Will Replace Them," AI adoption is expected to replace up to 70% of go-to-market (GTM) efforts in tech by 2025. This trend parallels historical shifts, such as the industrial revolution, where automation replaced labor-intensive tasks. Notably:
Human-driven GTM strategies are becoming obsolete as AI provides scalable and sustainable solutions at lower costs.
Industries like SaaS are witnessing near-instant scalability with AI, eliminating the need for prolonged hiring and training cycles.
2. Uneven Impact Across Occupations and Regions
An EY report emphasizes the uneven exposure to AI across occupations and regions:
Highly skilled roles (e.g., professionals, technicians) face significant AI exposure, particularly in pattern detection, judgment-based, and optimization tasks.
Regions with higher wages, like the USA and Western Europe, show greater economic feasibility for AI adoption. In contrast, regions like Sub-Saharan Africa face limited automation potential due to lower economic returns.
3. Job Transformation and Creation
Harvard Business School’s study on GitHub Copilot revealed how AI shifts the nature of work:
Developers using AI tools spent less time on core technical tasks and more on managerial responsibilities.
Lower-skilled individuals benefited the most, showing increased efficiency and focus on productive work.
This transformation could flatten organizational hierarchies, making roles more autonomous and specialized.
4. Macroeconomic and Social Implications
An UK-focused report on AI’s labor market impact highlighted:
AI could save 25% of private-sector workforce time, equivalent to six million jobs annually.
While AI is projected to displace 1-3 million jobs in the UK, it could also generate new roles through economic growth and the emergence of novel industries.
Education and healthcare sectors could see productivity boosts, improving workforce quality and inclusivity.
Controversial Findings
There are several report and forecasts on the potential impact of AI on the job market. However, findings are controversial.
Displacement vs. Augmentation: While some argue AI will replace human labor, others highlight its potential to augment roles, enabling workers to focus on higher-value tasks.
Long-Term Economic Gains vs. Short-Term Disruptions: Immediate job losses contrast with optimistic projections of long-term economic growth, prompting debates on balancing policy interventions.
I tend to be an AI optimistic, in the sense that AI as a technology will help humanity to solve our most concerning problems, including climate change, hunger, environmental protection, and diseases. And I am also optimistic, that we will find out, what to do, when AI takes over most of our jobs. And it will.
If we take a look at the 2024 layoff trends, it becomes clear, that the most advanced companies, pioneering AI and automation, are the forerunners in layoffs as well. They use this new technology to automate white collar and expert jobs, even in such growing fields like cyber security.
I'm pretty sure that the coming years will bring massive transformation on the job market, and that will only be intensified by the rise of robotics an autonomous machines. IT has started at the top companies, with the best US firms like Tesla and co. It will continue in the large multinational companies in the coming years, followed by smaller, regional or local firms in the 2030s. Then, as usual late followers, the public sector will join as well.
Hey, wait, not necessarily: the Government in the US might be the first mover, thanks to Elon Musk and his Head of DOGE role, with a clear mission to reduce government spending by billions and billions of US dollars. He WILL use AI and automation for it, building digital twins for the government infrastructure, using Palantir's technology platform, and transform the way government and public services are done. Let's see, what he can achieve in the bureaucracy.
Conclusions and Recommendations
AI’s impact on jobs is multifaceted, offering both challenges and opportunities:
Reskilling and Education: Governments and corporations must invest in upskilling programs to prepare workers for AI-enhanced roles. At least for the upcoming decades. Then, we'll see. Maybe we all become artists and entertainers.
Policy Frameworks: Proactive policies are essential to mitigate displacement risks and address inequality. A universal basic income might be one of the solutions - but alone it won't be enough.
AI Regulation: Clear guidelines for ethical AI deployment can prevent biases and ensure equitable benefits. Keeping humans in the loop can prevent loosing complete control over high risk processes - and prevent valuable jobs for the future.
While AI poses significant challenges, its potential to enhance productivity and create new economic opportunities makes it a critical area for strategic focus. A balanced approach—emphasizing technological advancement alongside workforce adaptability—will determine whether AI becomes a force for human-centred progress or disruption.
AI Tools We love
ChatGPT now understands real-time video.
On last Thursday during a livestream, the company said that Advanced Voice Mode, its human-like conversational feature for ChatGPT, is getting vision. Using the ChatGPT app, users subscribed to ChatGPT Plus, Team, or Pro can point their phones at objects and have ChatGPT respond in near real time. (Unfortunately, this feature is not yet available in Europe. But it will come soon, as always.)
Advanced Voice Mode with vision can also understand what’s on a device’s screen via screen sharing. It can explain various settings menus, for example, or give suggestions on a math problem.
To access Advanced Voice Mode with vision, simply tap the voice icon next to the ChatGPT chat bar, then tap the video icon on the bottom left, which will start video. To screen-share, tap the three-dot menu and select “Share Screen.”
Rivals like Google and Meta are working on similar capabilities for their respective chatbot products.
What does it mean for you? Well, 2025 will be the year, when we get flooded with super intelligent AI assistants, being able not just to talk, but to hear and see everything. They will become our everyday assistants. Can't wait!
Culture Corner
"You created them to be your friends? Or your slaves?"
Movie: Ex Machina (2015)
Speaker: Ava
Why it’s iconic: This question forces viewers to confront the ethical implications of creating sentient AI and how we treat them.
Ava’s haunting question in Ex Machina (2015) resonates with growing concerns in AI development today. As artificial intelligence advances, the line between tool and sentient being blurs, forcing humanity to confront profound ethical dilemmas.
A recent study warns of the potential to unintentionally create conscious AI systems capable of experiencing suffering. This parallels the moral reckoning Ava provokes: What happens when creators imbue machines with the ability to think, feel, or even question their existence? Will they become partners in shaping the future, or mere instruments bound to our will?
Experts advocate for precautionary measures to prevent the inadvertent creation of sentient AI while urging us to plan for the welfare of such systems should they emerge. Like animal welfare, the responsibility lies not just in preventing harm but in acknowledging the rights of those capable of experiencing it.
Ava’s words, coupled with this real-world warning, challenge us to rethink what it means to create life. Are we prepared for the ethical responsibilities that come with it? Or will we repeat the cycles of domination that have marked human history? As we push the boundaries of technology, the question lingers: Are we building allies—or oppressors? The answer will define our very future and the essence of humanity itself.
AI Art
Optimus Awards organises daily AI Art Competitions. If you'd like to try out yourself (hmm... your prompting skills), then give it a chance! You can even win 100 bucks every single day!
Take a look at some of the artworks that have been created by enthusiastic members of this growing community.
Quote for You
“The real question is, when will we draft an artificial intelligence bill of rights? What will that consist of? And who will get to decide that?” —Gray Scott
Gray Scott is visionary thinker, futurist, artist, and philosopher, an avid commentator on the future of artificial intelligence, AI ethics, the singularity, techno-philosophy, and the transformation of human society.
I hope, you liked this piece.
Thanks for reading!
✌️Until next time!
Zoltan
👉 In case you missed something, check out my latest articles:
The Truth about AI Development's Slowdown - and Why You Shouldn't Care
Tech's Diverging Future Paths After the US Presidential Elections
🔔 Follow me on LinkedIn and subscribe to receive regular updates on AI, tech and digital transformation topics for managers and leaders.
🚀 Join me on Substack for the full experience, including the archive, deep dives, resources and podcasts — for free! 🚀