Author

Michael

What is Microsoft 365 Copilot

Microsoft Copilot is a Large Language Model utilized as a personal assistant that is integrated in Microsoft Apps like Word, Excel, PowerPoint, Outlook and Teams. Furthermore Microsoft will introduce a new Experience called “Business Chat”.

Copilot is the result of a consequent development of personal assistants for Microsoft Products. Many of you might remember “Clippy” the advice giving paper clip build into early Microsoft Office products. “Cortana” was the next personal assistant by Microsoft installed in various products. Copilot covers the same need but more.

The promise is to unleash creativity, unlock productivity and uplevel skills. Or will it destroy creativity? It will certainly produce vast amounts of data.

How does Copilot work

By beeing integrated in the most used business Apps Microsoft Copilot will change the work of millions of people.

Word: Copilot can integrate information from shared documents within the organization, and is able to generate an initial draft based on a given prompt. In addition, it has the ability to adjust the tone of the text to suit the intended audience, whether it be formal or informal.

Excel: Copilot can interpret natural language prompts to provide answers and recommendations, minimizing the need for complex formulas. It can produce various visualization and charts without altering the spreadsheet.

PowerPoint: Copilot can generate a slide deck based on an outline or a Word document, and simplify the creating of professional presentations.

Outlook: Copilot can not only respond quickly to routine emails, but also gather content from email threads and other Microsoft 365 sources. It can also adjust the tone and length of responses.

Teams: Copilot is capable of setting meeting agendas and schedules in Teams, and acting as an effective decision-maker by listing the pros and cons of any discussion and suggesting next steps.

Business Chat: Copilot enables users to access their calendar, emails, chats, documents, meetings, and contacts all in one place, facilitating collaboration and streamlining project management.

A whole new way to work

It is easy to see, that Microsoft Copilot will transform our understanding of work. Summarizing meetings and pointing out the keyfacts. Consider it done! You want a powerpoint presentation about the clients subject? Here you have it. Microsoft is giving every office employee a powerful personal assistant that will increase the capacity, throughput and efficiency of it’s users.

Hatespeech in your proposal deck

As impressive as Copilot is, using the generated output of it unfiltered can lead to problems. Large language models can fail by inserting e.g. gender or racial bias into text or make things up by hallucinating. It’s alarming enough that those quirks can be found in output of search engines but imagine the consequences of such biases in Excel spreadsheets or your email inbox.

Here we also touch another important subject this blog will try to dive deeper in. Who is responsible if Copilot insulted your Client? Should the Large Language Model be blamed or should the user that didn’t quality check the output be blamed?

The value of a handwritten letter

Since Copilot will certainly boost the generation of Documents, Presentations, E-Mails etc. we must ask who will still be able to intake all that data ? Certainly this flood will increase the need of prioritizing your informational intake. Considering this, how much would you appreciate a handwritten letter?

April 28, 2023 0 comment 946 views

RESEARCH

The false hope of current approaches to explainable artificial intelligence in health care

GoGo Gadget: ChatPDF

Summarize the Research Paper "The false hope of current approaches to explainable artificial intelligence in health care"

The research paper "The false hope of current approaches to explainable artificial intelligence in health care" argues that the current methods of explainable AI are unlikely to achieve the goals of patient-level decision support. The authors suggest that while AI will have a significant impact on medicine in the coming decades, it is important to ensure that this technology is implemented in a way that maximizes patient benefit. However, despite its intuitive appeal, explainability for patient-level decision making is unlikely to advance these goals in meaningful ways. The paper recommends that healthcare workers exercise appropriate caution when using explanations from an AI system and urges regulators to be judicious in listing explanations among the requirements needed for clinical deployment of AI.

Why this paper was selected

This paper has been selected due to following factors:

The paper provides an overview of approaches towards explainable AI
Health Care and Medicine are high-stake areas that can’t rely on approximations
The authors outline possible failures of explainable AI approaches
As a solution rigorous and thorough validation procedures are adviced

Explainability of individual or local decisions by AI

The authors discuss techniques for explainability that can generate general descriptions of how AI systems operate, but they note that these methods have limitations and can only offer unreliable or shallow explanations for specific decisions made by AI.

Inherent vs post-hoc explainability

Explainable AI approaches are categorized into two groups: inherent and post-hoc explainability:

Inherent explainability can be used for machine learning models that use a finite amount of input data with limited complexity in which the relatonship between input arguments can be clearly quantified and understood.

Post-hoc explainablity on the other hand is beeing used for modern AI systems that have high-dimensional and complex data and models in which the relationship between inputs and outputs cannot be quantified anymore.

Approaches towards AI explainability

For AI systems that give output on the input of images saliency maps can be used as a heatmap that reflect the degree of importance in an area on the image for the given output.

Interpretability gaps

The interpretability gap of explainability methods relies on humans to decide what a given explanation might mean. Unfortunately, the human tendency is to ascribe a positive interpretation: we assume that the feature we would find important is the one that was used (this is an example of a famously harmful cognitive error called confirmation bias). This problem is well summarised by computer scientist Cynthia Rudin: “You could have many explanations for what a complex model is doing. Do you just pick the one you ‘want’ to be correct?

Reasonability of decisions by AI

In the example of heat maps, the important question for users trying to understand an individual decision is not where the model was looking but instead whether it was reasonable that the model was looking in this region.

Sources

Ghassemi M., Oakden-Rayner L., Beam L.

The false hope of current approaches to explainable artificial intelligence in health care. The Lancet – 2019

April 23, 2023 0 comment 1.6K views

The key to making AI trustworthy is to ensure that it is provably aligned with human values. This means designing systems that are motivated to pursue our objectives rather than their own, that are transparent about their decision-making processes, and that can be easily audited and corrected when they make mistakes.

Stuart Russell - 2019

Hello world!

2 min read

GoGo Gadget: ChatGPT

Write a Hello World article for a blog with the domain "ai-in-control.com"

Hello and welcome to AI-In-Control, a blog dedicated to exploring the fascinating world of artificial intelligence and its role in shaping the future of our world.

In this blog, we aim to provide insights and updates on the latest developments in the field of AI and how it is being utilized to control various aspects of our lives. From self-driving cars to smart homes, from healthcare to education, AI is rapidly transforming the way we live and work.

As we move towards an increasingly automated world, it is essential to understand the opportunities and challenges that come with AI. We believe that AI can be a powerful tool for good, but it is crucial to ensure that it is designed and used responsibly.

At AI-In-Control, we will explore the ethical implications of AI and its impact on society. We will discuss the need for transparency, accountability, and regulation in the development and deployment of AI systems. We will also highlight the latest research in the field and its practical applications.

We invite you to join us on this exciting journey and learn more about AI's potential to shape our future. Our goal is to provide valuable insights and resources that will help you stay informed and engaged in this rapidly evolving field.

Thank you for visiting AI-In-Control, and we look forward to your feedback and participation in our discussions.

Stay tuned for more exciting content coming your way soon!

Who controls who?

How can we possibly try to control what we don’t understand? Let’s look at the above example. ChatGPT wrote a very good Hello World post for this blog. By looking in closer one suspicious phrase came to my attention:

“[…] how it [AI] is being utilized to control various aspects of our lives.”

The only context ChatGPT was given as an input was the domain name “ai-in-control.com” of the blog. Apparently in this section ChatGPT interpreted the domain name in a way that the blog is about how AI can be used to control us humans.

Obviously that’s not the goal of this blog. The goal is to aggregate and discuss current research on Artificial Intelligence with a focus on the topics of Explainable and Trustworthy AI. Among other questions we will dive deep into the following:

Why do we need to be responsible with AI?
How can we trust Artificial Intelligence?
Can the output of Artificial Intelligence be explainable?
What do regulators demand of Artificial Intelligence?
Where should we have a “human in the loop”?
Which ethical aspects should be considered?
How can Artificial Intelligence be tested?
What Artificial Intelligence Solutions do exist and how can you use them?

So buckle up, stay tuned and get your coffees ready!

April 20, 2023 0 comment 4.5K views

Newer Posts

Older Posts