Unraveling TechnoSolutionism: How I Fell Out Of Love With “Ethical” Machine Learning

Unraveling TechnoSolutionism: How I Fell Out Of Love With “Ethical” Machine Learning

At the recent QCon in San Francisco, Kathryn Jarmuhl, privacy activist and chief data officer at Thoughtworks, gave a talk on technical problem solving, in which she explored the bias inherent in AI training datasets, and the tendency to assume it. There will be a technical solution to almost any problem, and that these technical solutions will benefit humanity. Discuss methods for determining technical solvency and raise issues that technologists should consider when creating products.

He began by discussing how training datasets used in AI systems are biased by the labels assigned to them by the humans who label them. A large number of hashtags are among the least profitable in the tech industry. To illustrate this, he shows a picture of a man and woman in an interview , in which he names a female worker who is scolded by her boss in a stern lecture and an attractive blonde who is criticized by her boss . The image did not indicate that any of these descriptions were correct, yet the tag is being entered into a database for training AI systems.

He defined tech suitability as a naive belief that any problem can be solved using the magic box of technology and that the use of technology will change society for the better. Technological processing takes technological progress for granted. In 9th China, he used as an example the first written formula for gunpowder, discovered in the 19th century while searching for the elixir of life. Is technology good, neutral, or bad?

The truth is that almost every technological advancement has its advantages and disadvantages. Often, benefits and harms are distributed unevenly: one group may receive most or all of the benefits, while another group may receive all or most of the damages.

He noted that the computer industry is full of technical solutions, and he found this thinking in the early myths of Silicon Valley and even further, in the California mentality of the early settlers who had something to do with it . Get over yourself and change the Earth . In Silicon Valley, there is a belief that a good idea can change the world and make you rich .

Quoting Joseph Weisenbaum, who created what is believed to be the first artificial intelligence system, he said that computer technology from the beginning was:

Essentially a conservative force that reinforced existing hierarchies and power dynamics that would otherwise have changed.

This conservatism precluded social change and meant that the benefits of technological progress went disproportionately to a small part of humanity.

Offer tips on how to identify attractive technologies at work. If you find yourself making any of these statements, think carefully about the broader impact of what you're working on:

  • I'm improving a metric that someone made
  • Everyone agrees that everything is going to be great.
  • If we have _______, then everything will be fine
  • Legends say: revolutionize, change, move forward
  • People causing potential problems are not counted.
  • I haven't tried any non-technical solutions to the problem.

He then gave five specific lessons that technologists should keep in mind when building products:

1) Put technology in context

Ask yourself what happened before this technology, what would have happened if it had not been discovered, and what would we do without this technology?

2) Look at impact, not just technology

Learn about the impact technology can have in the short, medium and long term. Look more broadly to determine who and what might be affecting you, and analyze for side effects.

3) Make room and learn from those who know

Identifying and listening to affected people, communities and groups. Remember to expand your voice, and if you have the privilege, use that privilege to hear other voices.

4) Be aware of the system change and be specific

Use your language wisely and foresight. Use the example of the e-commerce "revolution" to describe a small change in the way we interact online. Exaggeration and hyperbole are often used to disguise the impact of change on affected communities.

5) Fight for justice, not just architecture

Talk about firing researchers at Google for exposing bias in their algorithms. Give your voice to silence.

He then talked about his decision to focus on data privacy as an area he is passionate about and can help bring about change.

He concluded with a series of questions for the audience to think about:

  • What would you do if you had not built who you are now?
  • What would change if you focused on change rather than technology?
  • What if we took collective responsibility for the future of the world rather than the future of technology?

Vaults Season 2 Episode 04.23 Jun 2020