Pria™ is an Augmented Intelligence (A/I) Mentor
A/I = Artificial Intelligence + Human Intelligence
Meet Pria! A Generative AI teaching assistant programmed to help students, faculty, and researchers with their digital education, research, and career goals.
Today’s students require real-time answers at the moment-of-need, but teachers are often overloaded with content curation, one-on-one teaching, and research. Our personalized generative AI mentor bridges this teaching-learning gap by offering personalized, course-specific mentoring to students at the moment of need.
Pria plugs into your LMS (Learning Management System) and supplements traditional and online teaching with 24×7 support for course materials, coding, discovery, image creation, and research. The AI mentor uses natural language processing to understand questions and GPT-4 Turbo (from Open AI) to choose the best combination of 28 trusted, cited sources to discover answers.
Pria can be customized with a unique name, icon, and personality for every teacher. She can also provide students with personalized recommendations and resources to help improve learning outcomes. Available at the click of a button, Pria appears in every window of the LMS – or in a separate browser tab. She supports speech-to-text and text-to-speech, with integrated memory, so you can have seamless, natural conversations with her.
Powered by the Generative Intelligence Engine™
As Pria evolves, she learns patience and engagement. Under the hood, Pria is processing every request through the Generative Intelligence Engine (GIE) to emerge as a virtual mentor for education and research. By leveraging the intelligence of LLMs, the GIE provides a seamless integration between artificial intelligence and human.
This helps to actualize the idea of “augmented intelligence” =→ an iterative refinement of knowledge where both the human and AI participate.
As the AI engine develops to provide increasingly valuable information, human intelligence is propelled into a new world of innovation.
The GIE comprises a number of LLMs with long-term memory, custom conversation agents, prompt engineering, and an embedded database of Praxis hand-curated scientific knowledge. This allows for a seamless conversation; Pria can write you code, generate AI-images, search and summarize the scientific literature, and write you a zen kōan all in a single conversation!
I have been using Pria in my BS/MS/PhD level bioinformatics courses since she was born in Jan 2023. Pria answers questions as well as an Assistant Professor and is even an asset in my computational biology research lab. I say this with 3 decades of teaching/research experience at multiple research institutions (Vanderbilt, Emory, UGA, Clemson).
- Pria finds excellent overviews of any topic (+/- STEM with actual sources/links and no phony reference hallucinations). Pria also can ask the students questions to see if they understand.
- One of the most surprising aspects of Pria is that she is not a human. Students can ask her questions without embarrassment or anxiety. This is so important these days!!!
- Pria provides pre-peer review of all scientific publications (we don’t use her to edit, just point out limitations in our science).
- Pria helps new programmers understand their code (Python, R, P4, bash, etc) by adding comments to each line and pointing out syntax errors.
She gets better every day!
Alex Feltus, Professor of Genetics & Biochemistry, Clemson University
Pria is Powered by 28 Trusted Sources
OpenAI Logic Engine chooses the best Trusted Sources to Answer Each Request
Pria Integrates with Any Platform
Pria Embeds Seamlessly with any Platform via a Simple SDK
For general queries, Pria leverages the latest version of OpenAI’s GPT technology. This includes the most recent GPT-4 API release, optimized for realistic conversations and relevant information. The world’s knowledge is at your fingertips.
With a focus on leveraging open-source technologies, Pria utilizes a custom implementation of StabilityAI’s Stable Diffusion Model to generate novel AI images. This capability will grow over time as we expand Pria’s features and integrate custom machine learning data sets.
With the capability to perform internet search, Pria can get real-time information for any question! She can even summarize webpages and compile keywords.
Plus, Pria can tap into the
world’s largest library at Google Scholar. This is like having a personal search concierge at your beckon call anytime, anywhere. As Google gets smarter, so does Pria.
With any scientific question, Pria can query the peer-reviewed literature and return relevant details from research along with appropriate citations. This allows you to tap into 35 million scientific papers without having to spend your entire life reading.
Vector-embedding storage with Pinecone allows Pria to have a real long-term memory, remembering your conversations and growing smarter with time. The more time you spend engaging with Pria, the better she will become at predicting and understanding your needs.
With help from Wolfram Alpha, Pria’s brain is a supercomputer. She can compute answers using Wolfram’s breakthrough technology & knowledgebases, relied on by millions of students & professionals.
Foundational Knowledge in AI Course
At the Governor’s School for Science and Math, Pria used LLM to help students learn the principles they needed to build their own LLMs. Meta, right?
Coding Assistance in Medical Bioinformatics
At Clemson University, Pria assisted medical bioinformatics students conduct research and write code in pursuit of new discoveries in cancer genomics. See “Pria in Action” above.
Research Assistance in BioHackathons
In recent BioHackathons at the Praxis Academy, Pria assisted students, faculty, and researchers test their hypotheses against scientific research.
Prototype Digital Art for Game Design
Pria tapped into her creative side and helped students build colorful backgrounds, sprites, and graphical artifacts for video game design.
Pria in Action
Genes are intervals of DNA scattered across chromosomes. One particularly important gene is BRCA2 – a DNA repair gene in normal cells – increases the risk for breast and ovarian cancer when it is disrupted:
Before you can find gene mutations, you must pinpoint the gene’s address in the genome. Finding the chromosome, start and stop positions for a given gene often requires writing code to parse out these data from huge files. The first step in writing this code, involves finding a specific gene identifier to search for in the coordinate file, and then extracting a gene position from a gene coordinate “GFF3” file.
What is the ENSEMBL ID for BRCA2?
45%–69% of women who inherit a harmful BRCA2 variant will develop breast cancer by 70–80 years of age … and 11%–17% of women who inherit a harmful BRCA2 variant will develop ovarian cancer by 70–80 years of age. (Source: National Cancer Institute)
The next step in our genetic discovery story is to write some parsing code in Python.
Write python code to extract ENSG00000139618 chromosome, start, and stop coordinates from a GFF3 file.
gff3_file = sys.argv
with open(gff3_file) as gff3:
for line in gff3:
fields = line.rstrip(“\r\n”).split(“\t”)
if fields == “gene”:
attributes = fields.split(“;”)
for attribute in attributes:
gene_id = attribute.replace(“ID=gene:”, “”)
if gene_id == “ENSG00000139618”:
print(fields, fields, fields, sep=”\t”)
The final step in our BRCA2 adventure is to make a Python script and run it.
Write a Python script to return the chromosome, start, and stop coordinates for BRAC2.
python parse4.py Homo_sapiens.GRCh38.109.gff3.2
13 32315086 32400268
As a final verification, I manually posted these coordinates in the UCSC Human Genome Browser, and it was the correct position for BRCA2!
Check it out:
Now I have working Python code that will perform a complex bioinformatics task. This code can be adapted for any gene or other feature in the human genome. It’s coffee time!