Are AI Robots Close to Taking over Human Jobs?

AI is already threatening knowledge workers jobs, how close are they to replacing manual labor

Imagine walking into a spotless facility every morning, not because of an overnight cleaning crew but thanks to the silent, efficient work of janitorial robots powered by advanced artificial intelligence (AI). This vision may become a reality, but it also brings incredible opportunities and daunting challenges.

Partnership with our friend Shubham Saboo
UNWIND AI NEWSLETTER

Start your day with a 3-minute breakfast read and stay in the loop with the latest in the AI world! Unwind AI is your go-to source for parsing the AI signal from the noise, with a special focus on generative AI technologies. Subscribe now for free and get insightful updates delivered straight to your inbox. https://unwindai.substack.com/

This week, the internet was on fire with this video from AI robotics maker Figure. However, never underestimate the difference between a good demo and a real product.

While this demo looks impressive, it’s probably far from being a robot that can replace a human. Here’s why.

When learning and processing information, humans rely on various sensory inputs, with language and vision being two of the most prominent. However, the bandwidth of these inputs differs significantly, and understanding these differences is crucial for developing artificial intelligence (AI).

Written or spoken language has a relatively low bandwidth compared to vision. A person can read about 270 words per minute, translating to roughly 12 bytes per second. To put this into perspective, a modern large language model (LLM) is typically trained with around 20 trillion bytes of data. It would take a human approximately 100,000 years to read this amount of data, assuming they read for 12 hours a day.

In contrast, vision has a much higher bandwidth, with each of our two optical nerves carrying about 20 megabytes per second. This means that the data bandwidth of visual perception is roughly 1.6 million times higher than language's. To further illustrate this point, consider that a 4-year-old child has been awake for 16,000 hours, during which they have processed around 1 quadrillion bytes of visual data – 50 times more than the biggest LLMs trained on all the publicly available text on the internet.

This means there are significant implications for the development of Artificial General Intelligence (AGI):

  • Text is redundant, and visual signals in the optical nerves are even more redundant (despite being 100x compressed versions of the photoreceptor outputs in the retina). Redundancy in data is essential for self-supervised learning (SSL) to capture the structure of the data. The more redundancy, the better SSL performs. Both text and visual signals in the optical nerves are highly redundant, making them suitable for SSL.

  • Most human knowledge, and nearly all animal knowledge, stems from sensory experiences of the physical world. Language, while important, is secondary to these experiences.

  • To achieve human-level AI, machines must learn from high-bandwidth sensory inputs like vision. Without this crucial component, reaching human-level AI is highly unlikely.

It's worth noting that even without vision and audition, humans can still become quite intelligent. However, the sense of touch, which has a high bandwidth, plays a vital role in learning and development. Think about Helen Keller, who was deaf and blind but accomplished extraordinary things. She had the sense of touch and the capabilities to reason and plan, which LLMs don’t have today.

Chipotle Founder Launches Kernel: A Glimpse into the Future of AI-Powered Restaurants

Steve Ells, the entrepreneur behind Chipotle, has embarked on a new venture called Kernel. Kernel is a vegan takeout restaurant that heavily relies on robotics and automation. With only three human employees per location, Kernel aims to revolutionize the quick-serve restaurant industry by embracing AI and sustainability.

At Kernel, customers order meals online and are given a precise pickup time. Upon arrival, they use an app to open a locker containing their order. The food is prepped off-site at a central kitchen and delivered hourly by e-bike, with plans to use robots and autonomous vehicles for this process eventually.

Robots will aid employees in serving vegan bowls at Kernel

The heart of Kernel's operation is a robotic system that assembles orders, featuring a customized robotic arm from Kuka. This system minimizes the need for human labor, allowing Kernel to pay higher wages to its employees and provide them with more meaningful work.

As AI and robotics advance, Ells envisions further reducing labor in food delivery and preparation. He believes this platform is built for future technology, with the potential for droids to be used in self-driving vehicles for food delivery.

Kernel's innovative approach to restaurant operations showcases the growing trend of robotics and automation in the food industry. While challenges still exist, Ells is confident that Kernel will set a new standard for quick-serve restaurants in the AI and sustainability age.

Electric Sheep Unveils AI-Powered Robotic Mowers Set to Revolutionize Outdoor Maintenance

Electric Sheep, a California-based company at the forefront of AI and robotics in outdoor services, has introduced its latest innovation: AI-powered robotic mowers. On January 30, 2024, the company hosted a live demonstration to showcase its cutting-edge technology, which promises to transform how we maintain our lawns and outdoor spaces.

Electric Sheep wants to use robots to revolutionize lawn care

The robotic mowers, equipped with advanced AI, can autonomously identify mowable areas without requiring boundary wires or pre-programming. This feature allows the robots to work alongside landscape crews in various settings, from residential lawns to parks and campuses, boosting productivity and precision.

According to Nag Murty, co-founder and CEO of Electric Sheep, the machines possess a form of "common sense" that enables them to detect humans in their path. When the AI-powered mower recognizes a person standing before it, it pauses and turns around to avoid any potential collision. Additionally, the mower has the memory to return to the spot where it encountered the person and resume mowing once the area is clear.

Of course this is a very bounded context, it’s not a general solution for all sorts of lawn care including hedge clipping and trimming trees. It’s just mowing grass, but it’s an excellent initial use of automation.

Robots are Coming but Won’t Reach Human Capabilities for Quite Some Time

Robots plus AI are quite a way off from replacing humans. However, what they can do in the shorter term and what they can do to replace humans in the longer term are two different things. Today, the ideal use cases for AI are ones with a bounded context, such as Kernel and Electric Sheep. In the long term, AI robots will need to use multimodal data - video and audio as well as physical feedback (for spatial awareness) and NLP(for instructions) before they truly replace human beings.

Prompt of the Week: The 48 Laws of Power

I am a fan of Ryan Holiday, who does the Daily Stoic, a very popular newsletter on Stoic philosophy (I know that’s probably not what many would consider popular, but trust me).

I’ve skimmed this book numerous times and plan to listen to the audiobook on an upcoming road trip, but I liked what I’ve read. This inspired me. Here’s the overview and why it’s relevant to this prompt.

The "48 Laws of Power" is a book authored by Robert Greene, first published in 1998. It presents a distillation of 3,000 years of the history of power into 48 well-explained laws. The book draws from historical figures' tactics, triumphs, and failures and synthesizes the philosophies of Machiavelli, Sun Tzu, and Carl Von Clausewitz, among others. It's designed as a guide to amassing power, navigating the complex dynamics of social relationships, and avoiding the pitfalls of power dynamics in various contexts, including business, politics, and everyday life.

The 48 laws are categorized under several broad themes, each presenting a principle or strategy aimed at acquiring power, maintaining it, or defending oneself against the power plays of others. These laws cover aspects such as learning to keep others dependent on you (Law 11), learning the importance of reputation and how to protect it (Law 5), and the power of keeping one's intentions ambiguous (Law 3).

Critics of the book argue that its advice often leans towards the manipulative and amoral, suggesting that its strategies might foster unethical behavior and promote a cynical view of human interactions. Supporters, however, view it as a realistic examination of the world's operational framework, offering insights that can be applied ethically within competitive environments.

It has found a wide readership among people interested in politics, business, and personal strategy, including some high-profile entrepreneurs and celebrities who have cited its influence on their strategic thinking.

In application, it's crucial to approach the content with a discerning mind, considering the ethical implications of each law and how it aligns with one's personal values and professional integrity.

Using the Prompts

When I saw this LinkedIn post by Arman Liaghat, it inspired this set of prompts, which is a riff on his.

Be as specific as possible about your business, the relationships you focus on, and the outcomes you aim for. This specificity will help in generating more tailored and actionable advice.

Use these prompts to implement The 48 Laws of Power’s secrets into your daily life :

Never Outshine the Master, Win Through Your Actions, Play to People's Fantasies:

Using Robert Greene's 48 Laws of Power apply the following principles. Using the principles (1) Never outshine the master, (9) Win through your actions, not argument, and (32) Play to people's fantasies from Robert Greene's '48 Laws of Power,' I aim to enhance my [type of business] operations. My team includes [describe your team], and we serve clients like [describe your clients]. Focusing on my relationship with [choose one: team, clients, or direct reports], how can I apply these laws to achieve [specific goal]?"

Conceal Your Intentions, Get Others to Do the Work for You, Learn to Keep Others Dependent on You:

Using Robert Greene's 48 Laws of Power apply the following principles. I plan to implement the strategies (3) Conceal your intentions, (7) Get others to do the work for you but always take the credit, and (11) Learn to keep others dependent on you in my business, which is [describe your business]. Regarding my interactions with [choose a relationship, e.g., a team member or client], how can I use these laws to navigate [specific business goal or situation] successfully?"

Guard Your Reputation with Your Life, Never Appear Too Perfect:

Using Robert Greene's 48 Laws of Power apply the following principles. To protect and enhance my business's reputation, I am interested in applying the rules (5) Guard your reputation with your life and (46) Never appear too perfect from 'The 48 Laws of Power' by Robert Greene. My business operates in [describe sector] and frequently deals with [mention a scenario, like negotiations or customer interactions]. How can I leverage these laws to uphold and possibly improve my reputation, especially to benefit [partnerships, customer retention, etc.]?

Enter Action with Boldness, Master the Art of Timing:

Using Robert Greene's 48 Laws of Power apply the following principles. Considering the strategic importance of timing in business decisions, I'm focusing on laws (28) Enter action with boldness and (35) Master the art of timing from Greene's book for a critical decision about [describe decision]. My business is in the [industry] sector, facing conditions like [market conditions, competitive landscape]. How can I identify the best timing for this action? And once determined, how should I boldly implement this strategy to benefit [sales, growth, market presence]?"