AI Intermediate

Hardware

Energy consumption

• AI consumes energy in the training phase and in the operational phase.
• GPT-4 was trained 90 - 100 days on 25000 Nvidia GPUs
  • 3125 physical computers (server) are needed
  • A server uses 6.5KWh
  • 100 days * 24 hours = 2400 hours per server
  • 2400 hours * 6,5KWh = 15600 KWh
  • 15600 KWh * 3125 servers = 48 750 000 KWh
  • Is around 19 000 metric tons CO2 (in USA)
  • Around 890 000 trees - 5000 households (estimation)

Energy Consumption

• Training is not the same as production!
• We are looking for Green AI and improve hardware to have efficient power consumption without having a negative impact on the performance or accuracy of a model

Energy Consumption

* AI for business: What's wrong, and how to get it right | ZDNet

* Global survey: The state of AI | McKinsey

Now

2030

If we ask Siri: "What is the weather going to be like tomorrow?"

How is this calculated?

Speech-to-text

Text-to-speech

< 0.01g

< 0.01g

0.5g

Natural language

processing

0.5g

10g

Moore's Law

Microchips

• Microchips manufacturers think in nanometers when placing electronics. One nanometre is 0.0000001 centimeters.
• We are pushing the limits of the physical boundaries.

Von Neumann

• In order to calculate, a computer needs a CPU (Central Processing Unit) and memory.
• Invented in 1946 by Jon Von Neumann
• Separation of CPU and memory

Processor

Memory

Memory

Port

AI Accelerators

• GPU (Graphical Processing Unit)
  • Used for video games
  • Is faster, more performant but energy-wise not suitable for training AI
• Place the calculations and memory at the same place

Apple M1 Chip, Apple Silicon

Neuromorphic Computing

Quantum Computing

Quantum Mechanics

Quantum Mechanics

Quantum computing - Cases

In the cloud

• You can use computer power form Google and Microsoft datacenters
• Calculations are done via an internet connection
• We call this Infrastructure-As-A-Service (IAAS) 

So, why do we need to know this?

AI Intermediate

Automation

Technopanics

In the modern age, many worry that technology is harming our brains or eroding societal norms – but this fear is far from a new phenomenon, as this infographic reveals...

Is AI stealing our jobs?

• Write down 5 arguments why you've decided to say yes or no
• After 5 minutes we will start a discussion
  • Speaking one at a time
  • Listen actively
  • Try to support arguments by your own experience or by things we covered in class
• Debate for 10 to 15 minutes
• Teacher is the moderator:
  • Keeps the discussion on track
  • Encouraging students to share their viewpoints

Is AI stealing our jobs?

• The answer is nuanced
• There are 4 variables:
  • Context
  • Impact
  • Intended results
  • Goals

Is AI stealing our jobs?

Is AI stealing our jobs?

Yuval Noah Harari

Mustafa Suleyman

What does the AI revolution mean for our jobs?

AI Intermediate

Ethical AI

Slaughterbots

Slaugherbots

• The technology behind this is real
  • Flying automatically
  • Face recognition
• But... "Simply deactivate the battery or wait until the drone runs out of memory"?

Deep Fakes

"I will not use my medical knowledge to violate human rights and civil liberty, even under treat"

Oath of Hippocrates

Four Fundamental Principles

• Ethics by design
• Security by design
• Fairness by design
• Explainability by design

Asimov's Laws

Asimov defined three laws of robotics, which still influence thinking about AI and ethics. These laws were set out for the first time in 1942 and are direct inspiration for sci-movies like Terminator, Blade Runner, etc.

• First Law: A robot may not injure a human being or, through inaction, allow a human being to come to harm
• Second Law: A robot must obey the orders given it by human beings except where such orders would conflict with the first law

Asimov's Laws

• Third Law: A robot must protect its own existence as long as such protection does not conflict with the First or Second law. The robot is therefore at the service of humankind and must not harm it. 
• Zeroth Law: A robot may not harm humanity, or, by inaction, allow humanity to come to harm.

EU Ethics Guidelines

1. Above all, defend fundamental human rights; you will not reduce, restrict or mislead human autonomy.
2. You will ensure the robustness and safety of AI systems, and build in mechanisms to eliminate errors or inconsistencies at every stage.
3. You will ensure data privacy and keep control of data in the hands of citizens.
4. You will be transparant and traceable.

EU Ethics Guidelines

5. You will always give priority to diversity, non-discrimination and fairness and take into account human talent and needs.

6. You will increase social and environmental well-being and contribute positively to social change.

7. You will build in mechanisms to ensure the responsibility and liability of AI systems and their results.

But how? AI EU Act...