Designing Human-Centered Artificial Intelligence (part-1)

Tarek AlSawaf
6 min readFeb 17, 2020

--

Designing Human Centered AI — Talk @AUC, by Tarek Elsawaf

This article is a presentation summary that was given in DevFest Egypt under Google Community Egypt (GDG) on 28th, December 2019. It was for the first time in Egypt to be concerned with a subject related to the human-centered AI

by Tarek Elsawaf supported by desirm.

Why HCAI?

The AI is greatly developing these days. Many communities are being built up worldwide with injecting huge investments in AI field. I think all of you are aware of the initiative held by Google regarding the Machine Learning.

1.
Locally, the situation in Egypt is developing very fast. However, some people still see that it is early to speak about such issues, while these issues are the main focus in Europe, USA, Australia and Asia. So, it is our role to spread awareness about such issues, even if such issues are not to be applicable in the present time.

2.
Some of us may travel abroad and may start his own business there, so it is vital to him to be aware of the level reached by the world in this field.

3.
It is crucial for each one of us to be aware of the strategies and concepts that enable us to have our own philosophy that will augment our technical knowledge not just doing and delivery tasks!

4.
Anyone of use performs his own task according to his own view, depending on a certain reference to be sure that his task is correctly performed. But, amid the work pressure, deliveries, deadlines and clients’ needs, we may forget about our targeted users and its surrounding context.

5.
The conception of considering the human factor, as part of the problem solving or as part of the creative thinking, is the most important act as we will see later on.

Understand HCAI

Mob. Exchange Activity This is an activity to help us understand the core of the topic through the feeling.

Each one of us has his own smart phone. Could you please exchange your smart phones with each other? Now, everyone of use his classmate’s mobile phone. The question now is: can you open the phone’s gallery, or any of the installed applications? Can you browse the phone freely?!

Of course each one of you is anxious and irritable because of this procedure .

How do you feel now when your mobile phone with its private data and information is in the hands of your classmate? Of course you are so upset.

Why are you upset!!, although everything is well programmed and it is tailored for its user?!

Now, we should know that everyone of use has his own psychology, which needs to be understood. Such a psychology is the base upon which decisions are taken. Accordingly, you may fail to decide about a certain product without considering the users’ psychologies.

Designing HCAI

We will start by elaborating the topic’s title, i.e. “Designing Human Centered AI”. The title has 3 sections, i.e. (Design, Human Centered and AI). Each section has a different color in order to make it easy to differentiate between the three sections during the presentation.

So, we will start by understanding each section separately to reach the final picture.

Designing AIHC

Let’s say that, we need to design an AI product, but we need to have this product to be human-centered.

First off, Let’s backward a step and start the discussion by what is meant by the word “design”.

What’s design?

Historically, the word “design” is associated with art like fashion, architecture, graphics or even painting.

Design is a process including general rules, technologies and even tools. So, we can say that “design” includes three basic things, i.e. culture, way of thinking and the way of executing the design itself through search and toolboxes.

In other words, “design” is the rendering of intent ‘says jared spool’

The intent is the drive of the outcome. The intent is the answer of the questions starting by “why”.

To deliver a distinguished outcome, we should realize that our product is meaningful and delivers innovative solutions to its users.

What’s AI?

I’ll not stop here for a long time as I don’t like to speak about the technical point of view, but I’ll give some simple examples to highlight the core notion.

The AI is based on three pillars, i.e. Data Science, ML and AI. We can say that the AI represents the tool and the ML is the material.

  1. The Data Science produces insights.
  2. Machine Learning (ML) produces predictions

Until here AI doesn’t do it’s full work so in this step we can say that this is automated AI, but after AI comes, it could be augmented AI.

3. Finally, The AI takes produce actions.
it can stop or operate an operation according to the factors or patterns determined by the ML.

Tom and Gerry

We all know Tom & Gerry cartoon. In one of Tom and Gerry episodes, the lady who owns the house brought a robot to replace Tom. The developer of the robot wrote algorithms to enable the robot to learn and to decide in order to be able to chase Gerry, who was innovative in setting plans to enter the house. The robot learnt well how to foil each attempt made by Gerry to enter the house.

We can say that the developer made a metric and algorithms to the robot. The metric was about how to chase Gerry or any other mouse. This was very successful.

Then, an unexpected action took place.

Gerry brought a big number of rats and took them to the house. The robot’s task was to execute the metric by chasing all rats. The robot ruined the house and didn’t care about the context or the surrounding. Therefore.

When it comes to use, we should care about details

When we build up an AI or an ML project. We need to take care of the following:

  1. The Data Science produces insights — about people.
  2. Machine Learning (ML) produces predictions — for people to use.
  3. AI takes produce actions — to help people not to achieve the pre-set metric.

Stay tuned for part-2!
Update: PART 2

Thanks for reading!

If this hasn’t excited you yet, stay tuned for Part 2 which includes real and practical references and more clarifications 👋

HI! I’m Tarek Elsawaf (IBM Certified in human-centric AI in early 2019) from the desirm. If you enjoyed learning about this and would like to take a crack at solving some of the problems mentioned above, head to desirm.

--

--

Tarek AlSawaf

15 yrs ago, I combined business & UX expertise in human-centered design. Delivering solutions for global brands. Backed by IBM, IDEO & Product School certs