AI has the power to shape our world, and is increasingly becoming part of our daily lives - but what happens when it inherits society’s biases? This talk dives into the hidden bias and stereotyping in AI, from hiring algorithms to facial recognition systems.
We’ll learn how training data fuels these problems, and how AI can reinforce prejudiced beliefs in users, including real-world examples from tech giants like Amazon and LinkedIn. But it’s not all bad news - there are ways to fix it.
Get ready for eye-opening examples, shocking data, and real solutions to make AI fairer for all.
During the expert session, Antonia engages the audience through methods like a Q&A session, live demonstrations of AI applications, and interactive exercises using the Mentimeter app to discuss ethical concerns related to AI, such as biased training data and underrepresentation.
She also highlights positive uses of AI for marginalized communities, emphasizing the importance of involving the community in creating solutions and preventing bias through tools like bias audits. Antonia concludes by encouraging everyone to educate themselves about potential biases and make informed decisions to ensure the positive impact of AI on society.
Recognizing the positive applications of AI in areas such as accessibility, healthcare, education, financial inclusion, and social justice for marginalized communities is essential. However, it's crucial to design and implement these technologies ethically and inclusively.
Be aware that AI learns from historical data and can amplify or replicate existing biases. Do regular bias audits using diverse and ethical training data to prevent reinforcing unfair decisions and promote equality and safety.
While AI can provide numerous benefits, it's essential to remember the importance of human connection, especially in situations that require empathy or high degrees of manual skill. Provide users with choices depending on the context of the application.
In this expert session, Antonia delves into the intricacies of artificial intelligence (AI), discussing various types and their applications, as well as the ethical concerns surrounding their development and implementation. Antonia begins by introducing different types of AI systems, such as neural networks and machine learning, which includes deep learning models like convolutional neural networks (CNNs) for image recognition and large language models (LLMs) for natural language processing.
She also covers recurrent neural networks (RNNs), deep reinforcement learning, and generative adversarial networks (GANs).
Furthermore, Antonia discusses different levels of intelligence in AI, including artificial narrow intelligence (ANI), artificial general intelligence (AGI), and artificial super intelligence (ASI), and shares a brief history of AI development, mentioning milestones like the invention of digital computers, Alan Turing's Turing test, the Dartmouth Conference, and the creation of early AIs. The ethical concerns related to AI are a significant part of this expert session.
Antonia emphasizes the importance of addressing biased training data and underrepresentation in AI to prevent safety and equality issues. She provides examples of biased AI systems, such as Amazon's resume screening system, and discusses the environmental impact of AI, particularly large language models like ChatGPT that require significant energy for cooling.
Creative theft, job displacement, and misinformation are other ethical concerns Antonia touches upon. For instance, she mentions that AI can be used to generate false information that manipulates public opinion and discourse.
However, the expert session is not all about the potential downsides of AI. Antonia highlights positive applications of AI for marginalized communities, such as accessibility and assistive technologies, healthcare, education, financial inclusion, and social justice.
She emphasizes the importance of involving communities in creating solutions to address their needs.
Antonia discusses job tasks least likely to be replaced by AI due to human dexterity and empathy. For example, she uses a hairdresser's role as an example of a task requiring a high degree of manual skill and emotional connection with clients.
In response to questions from the audience, Antonia suggests providing users with a choice between generative AI chatbots and allowing them to choose their preferred interaction method depending on the context of the application. She also shares her personal experiences of frustrating interactions with unresponsive AI customer service agents and emphasizes the importance of human connection in such situations.
Finally, Antonia discusses the topic of bias audits in AI systems to prevent reinforcing biased decisions and promote diversity within teams and organizations. She encourages everyone to educate themselves about potential biases and make informed decisions for a better future.
In conclusion, this expert session offers a clear and balanced depiction of Antonia's insights into the different types of AI, their applications, ethical concerns, and positive impacts on marginalized communities.
From the survey every participant gets when the session ends, in their own words.
Fantastic session, thank you :)
Very engaging and knowledgeable, Antonia is a great speaker in my opinion, thank you.
This was a really informative talk and really loved the content. Lots of great facts and science backed information
The evolution of AI and how Bias plays it's part in AI, really interesting, thank you
I was aware of bias in AI and how training data has an impact but this session was useful in further extending that knowledge and how we can be mindful of that
Antonia was not only engaging but delivered the presentation so well it was easy to stay alert. Considering session was after lunch
Different problems that effect good AI decisions BTW: - speaker has a cute white dog, that did not look like AI at all; I doubted between A and D but choose the wrong one (A) - AI makes the mistake to create pictures and voices too perfect
Great talk, very informative! Loved getting a slide with sources used, I'll definitely be reading up on some of the things metioned in this talk.
Interactive. There is a live Q&A, demonstrations of AI tools as they run, and exercises through a polling app so the room can react to what it is seeing. People are asked to weigh in, not just listen.
No. It is about what these systems do and where they go wrong, not how they are built. No maths and no code. The examples are things people recognise: hiring tools, facial recognition, everyday search.
Yes, and arguably more. Once a tool is part of the working day, the risk stops being whether to use it and becomes what it quietly gets wrong. The session works through where bias enters the training data, and where the output reinforces what the user already believed.
It applies whichever assistant you have. The mechanism is the same across tools because it sits in the training data and in how people read the output, not in the vendor. The session names real cases at large technology companies rather than staying abstract.
There is a fix half. Bias audits, what to look for in training data, involving the people a system affects in designing it, and where a human decision has to stay in the loop. It ends on what to do, not on how bad it is.
She works in technology herself and consults on diversity and inclusion, so she sees both the systems and their effects. She also speaks about prompt engineering and AI adoption, which is why the session lands on the practical side rather than the philosophical one.
It is interactive, and the block above says exactly how for this session. Nobody has to speak up who does not want to: the chat and the polls carry most of it, and the questions can come in anonymously.
AI has the power to shape our world, and is increasingly becoming part of our daily lives - but what happens when it inherits society’s biases? This talk dives into the hidden bias and stereotyping in AI, from hiring algorithms to facial recognition systems. It is built for a general working audience: no prior knowledge is assumed, and people who already know the subject get the research and the practical side rather than an introduction.
Antonia Forster is one of the experts we auditioned ourselves, and we produce every session so we see how each one lands. Participants rate this session 4.5 out of 5, across 384 responses, and rate Antonia personally 4.7 out of 5.
It runs 45 min / 15 min: the talk, then live questions. All you do is pick a date. We brief Antonia, produce the session, host and moderate it, run the Q&A, and send you the report within an hour of the end.
A team, a department, an employee network or every single employee at once, with no cap and no per-seat price. It runs inside your own Microsoft Teams, Zoom or Google Meet, or on a branded page we build. No integration, no login, no company device or e-mail.
The session is delivered in English. You can add high accuracy live subtitles in 8, 16 or 24 languages, so everyone follows in their own. That is a paid option, priced by the number of languages you need.
No, and that is by design. There is nothing to install, nothing to integrate and no connection to your systems, so there is nothing for IT to validate: we join your own Teams, Zoom or Meet as a guest speaker, or we host a page and your people open a link. No software, no accounts, and no data of yours that we can reach. It is why hundreds of companies, among them some of the largest technology and financial firms in the world, were able to start within a day.
We never receive or hold personal data about your employees. Because the session runs in your own environment, attendance and identity stay with you, and what we report back is aggregated: numbers, ratings and survey answers, not people. That is also why IT approval usually takes minutes rather than weeks.
You keep the recording and can share it internally. You also get attendance over time, participant ratings for the content and the speaker, the survey answers, and the topics your people asked for next.
It is no secret that technology is having a diversity crisis. Despite initiatives to diversify, technology is still perceived as…
"When I came out of the closet I was blackmailed, threatened, even bribed.
AI is transforming our lives inside and outside of the workplace - but it is only as effective as the input we give it.
As AI becomes part of our everyday routines, this session shows how to use generative tools (e.g., ChatGPT) to think more…
At this point, the integration of AI in the workplace seems inevitable as many professionals are using it daily.
This session helps you stay up to date with the fast-moving world of generative AI.
“Netflix’s biggest competitor is sleep.”That statement captures the reality of today’s attention economy.
This session gives participants a practical and up-to-date foundation for using generative AI well, with a focus on how it can…
We’ve all seen how generative AI can write, draw, code and create. The next big shift however isn’t about creating more content…