Author: Sarah Mercado
Information Technology
Cloud Infrastructure Management
Software Programming
Software Programming
Pulumi Copilot
This is a conversational chat interface that uses LLMs to help users quickly execute a variety of cloud infrastructure management tasks. A user is able to converse with the AI through chat and ask questions about any topic related to their cloud infrastructure.
Pulumi covers use cases such as:
- Generating Infrastructure as Code as opposed to an engineer manually writing the code;
- Providing insights on the team’s Cloud Usage and identifying Cost-saving Opportunities as opposed to running a report, interpreting the data, and creating an action plan;
- Compliance and staying Secure, as opposed to potentially purchasing a separate product that can do these features;
- Debugging Cloud failures, as opposed to having an engineer execute the debug manually;
- Deep-diving Documentation, as opposed to a person having to sift through multiple sources of information manually.
Software Programming
Software Programming
Software Programming
Project CodeNet
The goal of this project is to provide the AI-for-Code research community with a large-scale, diverse, and high quality curated dataset to drive innovation in AI techniques. The intended use case is for AI to potentially learn and distinguish correct codes from problematic ones; eventually down the line, the intent of the project is to explore automatic code correction. This use of AI has the potential to rewrite and modernise legacy code.
GitHub Copilot
This is an AI coding assistant for engineers. This offers autocomplete-style suggestions as you code; analyses the context in the file you are editing, as well as related files, and offers suggestions from within your text editor. GitHub Copilot is trained on all languages that appear in public repositories.
The intent for this capability is to reduce coding and debugging time for software engineers especially for the repetitive/routine tasks.
Entertainment
Video, Animation, 3D Generation
Video, Animation, 3D Generation
Video, Animation, 3D Generation
Sora (OpenAI)
This has a text-to-video model that can create realistic and imaginative scenes from text prompts. This technology has recently been used to create an AI-generated music video, which as of this writing, can easily be distinguished as something created by AI. It is under this category of AI use that highlights the potentially malicious use of “deep-fake” media.
Unreal Engine (Epic Games)
While this company is better known in the video gaming industry, their product provides work optimisation tools to developers that allow them to cut down on development time. Time efficiency can be achieved by utilising their models for tooling animation, rendering, and physics. This is more than just reusing existing assets but allowing AI to fill in the gaps between an initial state to a target state e.g. animating movements such as a character walking in a 3D environment.
Social Media Content Generation
Video, Animation, 3D Generation
Video, Animation, 3D Generation
AI has also been specifically built to generate a post for a user provided a text prompt.
Sample tools: SocialBee, Radaar, and Semrush
Given a text prompt, these tools can generate descriptive text about a topic and also a highly relevant image to go with the narrative of the post. Social media managers and digital marketing personnel may find this useful since they would need to generate a large volume of posts regularly to maintain and increase engagement for the accounts they manage.
Image Generation
Written Content Generation
Image Generation
Stable Diffusion/Diffusion Art, OpenAI (DALL-E)
These AI tools and art have been popping up on the Internet and have been generating a lot of buzz in the art communities online due to alleged copyright infringement. AI generates new images from existing ones that are accessible on the Internet. The user provides detailed prompts that describe the image they want to see, which may include the art type and style e.g. of famous painters like Van Gogh.
Audio Generation
Written Content Generation
Image Generation
OpenAI Audio Models
The most popular use of AI-enabled audio generation tools is converting text to speech. The most common use case would be to transcribe speech into text and translate many languages into English, or even convert text into spoken audio. This can provide language accommodations to online services e.g. when users interact with chatbots on a company website.
Written Content Generation
Written Content Generation
Written Content Generation
AI-generated text content may be the most popular use case of GenAI. OpenAI’s ChatGPT has the capability to write poetry, song lyrics, essays etc. given a user text prompt. Grammarly has also provided live assistance in writing and also provides online paragraph rewriter services. On the other hand, there are also AI tools that detect plagiarism in essays e.g. AI Checker.
Research
AI is being used in the following areas:
- Earth Science Foundation Models (Climate Research)
- Medicine - Chemistry (discovering molecules to aid in medical therapy)
- Fine-turning Models (OpenAI) - allows users to create their own models with their training data not from scratch but by fine-tuning a base model offering
Industry Use Cases
- AI Chatbots
- Banking - personalised interactions e.g. inquiring about insights relevant to your spending
- Government - 24/7 citizen support e.g. reporting incidents e.g. public road issues that require repair
- Healthcare - empower patients to quickly get information regarding simple inquiries e.g. service coverage, relevant facilities based on location and services
- Insurance - provide immediate responses that are personalised and accurate e.g. resolving issues related to submitting claims on policies
- Retail - deliver highly personalised, contextually relevant shopping experiences
- Marketing and Sales
- FETA - an AI that can read technical documentation crammed with words and images from machinery manuals, sales catalogues, how-to guides.
Work
Misinformation
Education
The potential productivity gains are promising but the displacement of labour that may happen in parallel with consideration to organisational cost. There are brewing concerns over customers or individuals at large not being able to distinguish between chatting with an AI or a real person, which may bring to question how impersonal customer service will be in the age of AI.
Education
Misinformation
Education
The potential of receiving personalised learning from AI may bridge the gap for students with varying learning needs may it be in the pace, the means (text or audio-based), the language. AI may also offer simplified summaries to complex concepts. On the other hand, learners who overly rely on AI may undermine their capacity to work independently and to think critically.
Misinformation
Misinformation
Misinformation
GenAI has the potential to reduce access to the availability and the quality of information in the case where the fine-tuning of the model is influenced by malicious parties with the use of bots that have pre-defined text content/prompts.
Equality/Inequality
Equality/Inequality
Equality/Inequality
In the age of Generative AI, the party with the capability to host and process a significant volume of data (foundation model) earlier in the game for their business will have the advantage over those who adapt or utilise AI later. The sentiment is similar for the high-skilled individuals who have not been enabled or trained to work alongside AI - AI complementing human effort as opposed to replacing humans.
One concern that has been raised related to the rise of AI and the devaluation of high-skilled work is how people may eventually see little value in investing in a degree or certification for specific skills.
Healthcare
Equality/Inequality
Equality/Inequality
AI can potentially improve diagnostics and predictive capabilities, which can improve patient outcomes. An example would be using GenAI to read and analyse medical imaging e.g. X-rays or MRI scans. Another potential advantage is for AI to enable patients to manage their health more proactively through apps that are accessible outside of clinical settings.