Researching AI technologies by Elizabeth Georgiades

jeesivazquez — 2025-05-05 10:09:42 UTC

Risk: It is difficult to explain how AI works (it is not a step by step program that you can follow and understand how it got to a conclusion). I find that risky because we don't understand why it decided a specific thing (for example, "helping" judges to take decisions about people with an AI that was trained with bias data) and in consecuence it is hard to appeal.

How does research help: even if we cannot understand 100% how the AI makes decisions it's important to develop technology on top of the AI that helps us understand the probability of bias in X decision, if the datasets are bias and how to clean them. Also understand if there are prompts that minimize the risks of bias in the output.

2025-05-05 10:10:09 UTC

Healthcare is a field where AI could be really helpful. Analysing large datasets containing the medical history of a large number of patients (including also personal information), AI could find illness patterns and possible solutions. Due to the importance of the topic, prompting must be done by specialised personnel. A tool that could be used for this goal is Google Cloud BigQuery.

2025-05-05 10:19:47 UTC

To answer it, I’d start by using Gamma to create a draft of the presentation, I think it helps to organise the structure quickly using AI. Then I’d use Canva to make the slides more visual and clear since it also has AI tools for image creation. I’d also use ChatGPT to help me build and improve my arguments.

For example, AI can help analyse large amounts of data by finding patterns and trends that would take a person a lot of time to identify manually.

2025-05-05 10:28:43 UTC

The AI allows us to provide responses at a much faster rate and connect information much quicker. This makes the knowledge dissemination much quicker and frees the minds to focus on developing new innovations and concepts.