My new path in AI ethics + a new pilot + I’d love your help!

Hi everyone,

Most of you probably know me through my work in the fields of AI ethics. Since 2021, I’ve been consulting, writing papers, creating content, speaking, and conducting research on how to do AI responsibly.

In the last 6 months, I took a little pivot. I still see myself as doing AI ethics, but in a new way that I think will be more effective. You already know a bit about it if you read my last posts here. Today, I want to share why I decided to do this, because it comes from a reflection on the field of AI ethics. I would love your help, and I invite you to a pilot!

It’s been a wild ride in AI ethics over the last few years. In the early days, AI ethics came mostly from academia, then it started to drip into industry, and some organizations were genuinely excited about it.

Over time, I noticed some counterproductive trends:

1. A divide between ethics and building

AI ethics is typically communicated as a topic separate from building AI. One place we see it is in AI education. Often, if AI ethics topics are present at all, they appear at the end, as a unit that would be dropped if there isn’t enough time. This results in marginalizing AI ethics and seeing it as something that slows things down, gets in the way, and stops progress.

2. Emphasis on theory

A lot of the activity in AI ethics focuses on the theoretical side of things. The most common activity (as I observed in both my consulting and my research) is writing documents about what ought to be done, like AI ethics principles. These are helpful, for sure, but we have a gap between the theory and the practice. This has been a topic of conversation for a long time, and unfortunately, we are still not seeing enough progress.

3. Mostly directed at those who develop AI rather than those who use it

AI ethics content is mostly directed at organizations that develop AI, and especially at people involved in its production, like software engineers. There is also content for procurement and investment. I produced a lot of content of these types and it is important. But very little is aimed at supporting AI consumers in the general public. For example, what does it mean to prompt ethically or build agents ethically?

5. Emphasis on criticism rather than solutions

Journalists, academics, and activists shine a light on the negative consequences of AI. For example, one of the topics closest to my heart is how using generative AI can lead to cognitive decline. At this point, there are multiple research studies showing this, such as a recent MIT study showing how common ways of using generative AI can lead to reduced brain activity. It is very important to uncover these things. But there aren’t enough explanations on how to avoid such consequences.

6. Lack of alternatives for hype narratives

There are plenty of misleading narratives around AI. For example, that AI is human-like (an assistant, a coach, a partner) and that AI will replace people. Many in AI ethics, myself included, disagree with both. But there aren’t a lot of well-developed alternatives.

These observations made me realize that I want to do things differently. And that is what I started doing over the last six months.

My goals

• Unify building and ethics

• Support AI consumers in the general public

• Create new narratives and help people think critically about AI

What I created: The Chef Approach to AI

If you’ve been following my writing recently, you have already read about the approach that I developed to accomplish these goals. I call it the “chef approach to AI”. This is an approach for using generative AI ethically and thoughtfully to do meaningful work and even reinvent professions. The heart of it is understanding how to use AI to create processes rather than just using it to produce an output. It is like the difference between a chef who designs a recipe and owns the execution and a consumer ordering takeout. This way of approaching AI:

• Avoids cognitive decline because it doesn’t require offloading tasks to AI

• Avoids automating ourselves out because it leads to creating a new version of the work

• Creates an AI narrative that resists both the hype and fear-mongering

• Allows for easier alignment with values because it leaves more room for embedding values in tools that we build

• Typically produces better outputs because it incorporates the user’s expertise

Here is one example, a longer general explanation, and a webinar I gave about it.

Courses

I haven’t really shared this publicly until now, but I developed a new course based on this approach. I’ve done three pilots and now I am on my first paid course. In addition, I have just created an on-demand version (you can see all of them on the courses page on my website).

The courses bring building and ethics together:

• Building - the courses teach how to build custom chatbots

• Ethics and critical thinking - the courses show how to think critically about AI and how to embed values at every step of the way.

As you can see, I am starting something new. I still see myself as doing AI ethics, but in a new way that I think will be more effective. I am super excited about it! If you are excited about this work too, I would love your help:

Talks

I am looking for places to give more talks. Let me know if you have ideas or know someone who might be interested.

Networking

I want to start giving my talks and courses at organizations. I am at the customer discovery phase. I’d love to chat with people who work at organizations that have offered or are interested in offering generative AI training of the kind that I described. I want to understand what they need and what they have been doing so far. If this is you or you know someone who fits this description, please reach out.

Pilot

I am about to start another pilot! In this pilot, participants will go through the on-demand course at their own pace but with my support. They will be welcome to participate in up to three live virtual office hours.

I accept participants by application and it is only open to subscribers of this newsletter, the AI Treasure Chest. Paid subscribers are welcome to participate for free. For free subscribers, the cost is $47, which is the cost of the on-demand course itself, so the live sessions are free.

You can read more details and apply here:

Apply to the pilot

Reach out

If you’d like to help, please reach out to me by responding to this email (if you received this as an email) or through my contact page here. If you would like to apply for the pilot, use this application form.

An AI-generated take on this post!