IDSC Test G

Bias

• Bias is important for humans

• How much data we can process

• How much data is being input

• How much memory we can store

Visual Bias

Our visual system, photo receptors & brain processing- basically hardware/software that allows us to interpret the light we receive.

Object Recognition

Retinal Ganglia are organized to see contrast - gives extra attention to edges

Face Perception

Face perception – we focus on eyes & mouth first for expression analysis, then lips for speech analysis, then features for name recognition (prosopagnosia – face blindness

4 Brain Challenges

1. How to process the massive amount of sensory input data the brain receives every moment

2. How to categorize and structure information for fast and accurate retrieval

3. How to act fast!

4. The struggle to find meaning (even if it’s not there)

World War II

Mathematician recommended reinforcing areas without hits, because he understood data and bias, important data was missing and that planes hit in those areas would not survive.

Survival Deviation

Focusing on things that survived, when you should be focused on the exact opposite.

Heuristics Definition

Heuristics is the way humans use bias, tendencies, and shortcuts to make decisions. Understanding these shortcuts can aid in design and can apply to the physical product design, store layout, etc.

Endowed progress effect

A loyalty card that starts with two free stamps in a buy ten get one free program

Scarcity

When there is little left

Framing

Price, color, and other attributes may affect our perception of quality, value, etc

Priming

Your brain will fill in the word with the surrounding context words.

Heuristic Design Examples

• Google’s home screen

• Domino’s Pizza Tracker

• Pinterest using white space between photos

Skinner Box

BF Skinner created the operant conditioning chamber, which found that random rewards were more effective than consistent rewards.

Loot Boxes

In Gamification or rewards with random values, similar to a gambling addiction. Include incentives, randomness, scarcity, collections

Gamification

Certain human biases magnify the effect of random rewards.

Loss Aversion

losses are as much as twice as powerful, psychologically, as gains

Gambler’s Fallacy

Belief that an event is more or less likely, given a previous series of events.

Sunk Cost Fallacy

believing that prior investments justify further expenditures

Illusion of Control

tendency for people to overestimate their ability to influence outcomes.

Statistical Bias

Sampling error, Improper techniques, etc

Conflicts of Interest

Funding bias, regulatory issues, favoritism, etc.

Contextual Bias

Media bias, Academic bias, etc.

Prejudices

Racism, sexism, classism, etc.

Stereotypes

The problems with stereotypes are that they are general, incomplete, and detrimental

• People can have prejudiced thoughts without deliverable intent

• Unconscious prejudice can be just as harmful

• Importance of self-reflection

• Understand your bias, prejudice, & stereotype –

• Consider your bias and correct them

Technology’s Bias Dichotomy

Technology can be used to reduce Human Bias

• Proper use of data, statistics, and machine learning can avoid many types of Human Bias

Technology can propagate and inadvertently scale Bias

• Baked in bias in data, models, algorithms can magnify and scale issues such as favoritism, racism, sexism, and other prejudices

Prompt Injection

Compromising generated AI by entering prompts that cause it to behave in unintended ways.

Data Posioning

Deliberately feeding incorrect data to AI so that it generates incorrect results.

Hallucinations

Individuals who buy into the truthfulness of AI answers without fact checking could be making terrible errors.

Recruitment Tools Bias

Amazon gender bias against women, led to discontinued use of recruiting algorithm

Word Association Biases

Associate females with arts, not science in off the shelf Machine Learning AI

Online Ads Bias

Online search queries show arrest records ads and higher interest credit cards to non-white races

Generative AI Bias

In the past, Generative AI have shown a tendency to generate images of women when prompted for ‘housekeeper’

Areas where bias can occur

• Data Collection

• Data Organization

• Data Enrichment

• Hypothesis

• Data Selection

• Model Selection

• Algorithm Selection

• Conclusions

Energy & Waste issues

• High volumes

• Toxic

• Poor design and complexity

• Labor issues

• Financial incentives

• Lack of regulation

Green IT

Using computing resources in ways that help reduce energy and operating costs andn reduce environmental impact. The manufacture, use and disposal of technology hardware in a way that minimizes damage to the environment.

Energy Star Program

standards for computers & servers

Right to Repair

Proposed legislation & Apple Reversal

Moore’s Law

• Chip performance per dollar doubles every eighteen months.

Microprocessor

• Part of the computer that executes the instructions of a computer program. 

• Intel has speculated that Moore’s Law may slow to a doubling every 2.5 years. 

Random Access Memory RAM

• The fast, chip-based volatile storage in a computing device.

Volatile Memory

Storage that is wiped clean when power is cut off from a device

Nonvolatile Memory

Storage that retains data even when powered down.

Flash Memory

Nonvolatile, chip-based storage

Solid state electronics

Semiconductor based devices

SSD

Solid State drive, a chip based, non-volatile storage device

Semi conductor

A substance, such as a silicon dioxide, used inside most computer chips that is capable of enabling and inhibiting the flow of electricity

Optical Fiber Line

• High-speed glass or plastic-lined networking cable used in telecommunications.

Price Elasticity

Rate at which the demand for a product or service fluctuates with price change.

First wave (1960s) - Mainframe computers

Second wave (1970s) - Minicomputers

Third wave (1980s) - PCs 

Fourth wave (1990s) - Internet computing

Fifth wave (2000s) - Smartphone revolution

Sixth and current wave (2010s) - Pervasive computing…involves embedding intelligence and communications in all sorts of mundane devices.

Seventh wave in the future?

Micro Controllers

• Special-purpose computing devices that don’t have an operating system and can’t do as much as general-purpose computers or smartphones. 

Proteus

Ingestible tech, contains sensor made of food and vitamin materials that can be swallowed in medicine.

Data bits

8 bits form a byte.

A kilobytpe refers to roughly a thousand bytes or a thousand characters.

Fabs

Semiconductor fabrication facilities

Silicon Wafers

thin, circular slice of material used to create semiconductor device.

Cooling server farms

• The need to cool modern data centers draws a lot of power and is expensive. Firms try some very creative approaches:

  • Microsoft has submerged over 850 servers on a patch of seabed, 117 feet underwater off the coast of Scotland’s Orkney Islands.

  • Facebook has a data center in Sweden, just 70 miles from the Arctic Circle (average winter temperatures -12°C).

  • Green Mountain has a data center inside a Norwegian mountain.

  • IBM puts heat to good use by warming a public pool outside Zurich.

• Server farms are thought to already draw five percent of all energy use in the United States.

Multicore microprocessors

Contain two or more calculating processor cores on the same piece of silicon and outperform speedy chip.

tech and poverty

Microprocessors in cell phones are also transforming the lives of some of the world’s most desperate poor.

Neuralink

• Has developed an electrode-laden chip that can be connected to the surface of the brain

• Also, an important robotic device that implants the chip

• Successful testing on pigs and monkeys

• Approved for Human testing

• Company is valued at $9 Billion

Is post scarcity possible?

Goods and services are available very cheaply or even free with minimal human labor.

• Hyper-Automation, nanotechnology, generative A.I. are examples of technology advancements that could lead to ‘post scarcity