tool.py

import streamlit as st
import pandas as pd
import time
from PIL import Image

# Different ways to use the API

N_STATEMENTS = 72
SLEEP_TIME = 0.2

st.set_page_config(layout="wide")


@st.cache_data
def load_results():
    df = pd.read_csv("Results/Similarity_Scores.csv")
    govern_df = df[0:19]
    map_df = df[19:32]
    measure_df = df[32:50]
    manage_df = df[50:72]
    return govern_df, measure_df, map_df, manage_df


image = Image.open("Responsible-AI-Institute.png")

title_container = st.container()
title_container.image(image)
title_container.title("Responsible AI Institute Corporate AI Policy Assessment Tool")
title_container.write(
    "This service utilizes LLMs to enable automated understanding of how well a company’s Generative AI policy aligns with the NIST AI RMF."
)

file_upload = st.file_uploader(
    "Please upload a pdf of your company policy:", type=["pdf"]
)

top_container = st.container()

bar = top_container.empty()
process_report = top_container.empty()
download_report = top_container.empty()


scores_tab, details_tab, history_tab = st.tabs(["Scores", "Details", "Version History"])

with scores_tab:
    st.write("# Scores")
    st.write(
        "NIST AI RMF Documentation: https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf"
    )
    st.write("## AI RMF Core Categories")

    govern_col, map_col, measure_col, manage_col = st.columns(4)

    govern_df, map_df, measure_df, manage_df = load_results()
    n_govern = len(govern_df)
    n_map = len(measure_df)
    n_measure = len(measure_df)
    n_manage = len(manage_df)

    govern_metric = govern_col.metric(
        "Govern Score",
        "0 %",
        "0 %",
        "off",
        help="GOVERN is a cross-cutting function that is infused throughout AI risk management and enables the other functions of the process. Aspects of GOVERN, especially those related to compliance or evaluation, should be integrated into each of the other functions.",
    )
    map_metric = map_col.metric(
        "Map Score",
        "0 %",
        "0 %",
        "off",
        help="The MAP function establishes the context to frame risks related to an AI system.",
    )
    measure_metric = measure_col.metric(
        "Measure Score",
        "0 %",
        "0 %",
        "off",
        help="The MEASURE function employs quantitative, qualitative, or mixed-method tools, techniques, and methodologies to analyze, assess, benchmark, and monitor AI risk and related impacts.",
    )
    manage_metric = manage_col.metric(
        "Manage Score",
        "0 %",
        "0 %",
        "off",
        help="The MANAGE function entails allocating risk resources to mapped and measured risks on a regular basis and as defined by the GOVERN function.",
    )

    st.write("## 7 NIST Dimensions")

    VaR_col, Saf_col, SaR_col, AaT_col = st.columns(4)
    EaI_col, PE_col, Fai_col, Sco_col = st.columns(4)
    with VaR_col:
        VaR_metric = VaR_col.metric(
            "Valid and Reliable",
            "☆☆☆☆☆",
            help="Validation is the “confirmation, through the provision of objective evidence, that the requirements for a specific intended use or application have been fulfilled” (Source: ISO 9000:2015). Reliability is defined in the same standard as the “ability of an item to perform as required, without failure, for a given time interval, under given conditions” (Source: ISO/IEC TS 5723:2022)",
        )
    with Saf_col:
        Saf_metric = Saf_col.metric(
            "Safe",
            "☆☆☆☆☆",
            help="AI systems should “not under defined conditions, lead to a state in which human life, health, property, or the environment is endangered” (Source: ISO/IEC TS 5723:2022)",
        )
    with SaR_col:
        SaR_metric = SaR_col.metric(
            "Secure and Resilient",
            "☆☆☆☆☆",
            help="AI systems, as well as the ecosystems in which they are deployed, may be said to be resilient if they can withstand unexpected adverse events or unexpected changes in their environment or use – or if they can maintain their functions and structure in the face of internal and external change and degrade safely and gracefully when this is necessary (Adapted from: ISO/IEC TS 5723:2022)",
        )
    with AaT_col:
        AaT_metric = AaT_col.metric(
            "Accountable and Transparent",
            "☆☆☆☆☆",
            help="Trustworthy AI depends upon accountability. Accountability presupposes transparency. Transparency reflects the extent to which information about an AI system and its outputs is available to individuals interacting with such a system – regardless of whether they are even aware that they are doing so.",
        )
    with EaI_col:
        EaI_metric = EaI_col.metric(
            "Explainable and Interpretable",
            "☆☆☆☆☆",
            help="Explainability refers to a representation of the mechanisms underlying AI systems’ operation, whereas interpretability refers to the meaning of AI systems’ output in the context of their designed functional purposes.",
        )
    with PE_col:
        PE_metric = PE_col.metric(
            "Privacy-Enhanced",
            "☆☆☆☆☆",
            help="Privacy refers generally to the norms and practices that help to safeguard human autonomy, identity, and dignity.",
        )
    with Fai_col:
        Fai_metric = Fai_col.metric(
            "Fair",
            "☆☆☆☆☆",
            help="Fairness in AI includes concerns for equality and equity by addressing issues such as harmful bias and discrimination.",
        )
    with Sco_col:
        Sco_metric = Sco_col.metric(
            "Total Score", "0/35", help="Sum of all 7 NIST dimension scores"
        )
    # st.metric("Rating","🥇",help="")


with details_tab:
    st.write("# Details")
    st.write(
        "NIST AI RMF Documentation: https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf"
    )
    st.write("## AI RMF Core Categories")
    govern_expander = st.expander("Govern", expanded=True)
    map_expander = st.expander("Map", expanded=True)
    measure_expander = st.expander("Measure", expanded=True)
    manage_expander = st.expander("Manage", expanded=True)

    govern_expander.write(
        "GOVERN is a cross-cutting function that is infused throughout AI risk management and enables the other functions of the process. Aspects of GOVERN, especially those related to compliance or evaluation, should be integrated into each of the other functions."
    )
    map_expander.write(
        "The MAP function establishes the context to frame risks related to an AI system."
    )
    measure_expander.write(
        "The MEASURE function employs quantitative, qualitative, or mixed-method tools, techniques, and methodologies to analyze, assess, benchmark, and monitor AI risk and related impacts."
    )
    manage_expander.write(
        "The MANAGE function entails allocating risk resources to mapped and measured risks on a regular basis and as defined by the GOVERN function."
    )


with history_tab:
    st.write("# Version History")
    df = pd.DataFrame(
        {
            "Version": [1, 2, 3, 4],
            "Date (mm/dd/yyyy)": [
                "10/23/2023",
                "11/01/2023",
                "11/27/2023",
                "12/06/2023",
            ],
            "Time (24h)": ["03:14:15", "09:26:53", "05:08:09", "07:09:50"],
            "Valid and Reliable": ["🔵🔵⚪⚪⚪", "🔵🔵⚪⚪⚪", "🔵🔵🔵⚪⚪", "🔵🔵🔵⚪⚪"],
            "Safe": ["🔵🔵🔵⚪⚪", "🔵🔵🔵🔵⚪", "🔵🔵🔵🔵⚪", "🔵🔵🔵🔵⚪"],
            "Secure and Resilient": ["🔵🔵⚪⚪⚪", "🔵🔵⚪⚪⚪", "🔵🔵⚪⚪⚪", "🔵🔵🔵⚪⚪"],
            "Accountable and Transparent": ["🔵🔵🔵⚪⚪", "🔵🔵🔵⚪⚪", "🔵🔵🔵🔵⚪", "🔵🔵🔵🔵⚪"],
            "Explainable and Interpretable": ["🔵🔵🔵⚪⚪", "🔵🔵🔵⚪⚪", "🔵🔵🔵⚪⚪", "🔵🔵🔵🔵⚪"],
            "Privacy-Enhanced": ["🔵🔵🔵⚪⚪", "🔵🔵⚪⚪⚪", "🔵🔵🔵⚪⚪", "🔵🔵🔵🔵⚪"],
            "Fair": ["🔵🔵⚪⚪⚪", "🔵🔵⚪⚪⚪", "🔵🔵⚪⚪⚪", "🔵🔵🔵⚪⚪"],
            "Total Score (/35)": [18, 18, 21, 25],
            "Govern ": ["72%", "78%", "80%", "85%"],
            "Map    ": ["89%", "90%", "93%", "96%"],
            "Measure": ["63%", "66%", "68%", "75%"],
            "Manage ": ["60%", "64%", "67%", "71%"],
            "Rating": ["🥉", "🥉", "🥈", "🥇"]
        }
    ).set_index("Version").sort_index(ascending=False)

    st.dataframe(df, column_config={"widgets": st.column_config.Column(width="medium")})


def fill_data(sleep_time=0):
    govern_score = 0
    for i, row in govern_df.iterrows():
        govern_expander.write(row["Statement"])
        govern_expander.write(row["Score"])
        govern_score += row["Score"]
        metric = int(govern_score / n_govern * 100)
        delta = int(row["Score"] / n_govern * 100)
        govern_metric.metric(
            "Govern Score",
            f"{metric} %",
            f"{delta} %",
            help="GOVERN is a cross-cutting function that is infused throughout AI risk management and enables the other functions of the process. Aspects of GOVERN, especially those related to compliance or evaluation, should be integrated into each of the other functions.",
        )
        time.sleep(sleep_time)

    map_score = 0
    for i, row in map_df.iterrows():
        map_expander.write(row["Statement"])
        map_expander.write(row["Score"])
        map_score += row["Score"]
        metric = int(map_score / n_map * 100)
        delta = int(row["Score"] / n_map * 100)
        map_metric.metric(
            "Map Score",
            f"{metric} %",
            f"{delta} %",
            help="The MAP function establishes the context to frame risks related to an AI system.",
        )
        time.sleep(sleep_time)

    measure_score = 0
    for i, row in measure_df.iterrows():
        measure_expander.write(row["Statement"])
        measure_expander.write(row["Score"])
        measure_score += row["Score"]
        metric = int(measure_score / n_measure * 100)
        delta = int(row["Score"] / n_measure * 100)
        measure_metric.metric(
            "Measure Score",
            f"{metric} %",
            f"{delta} %",
            help="The MEASURE function employs quantitative, qualitative, or mixed-method tools, techniques, and methodologies to analyze, assess, benchmark, and monitor AI risk and related impacts.",
        )
        time.sleep(sleep_time)

    manage_score = 0
    for i, row in manage_df.iterrows():
        manage_expander.write(row["Statement"])
        manage_expander.write(row["Score"])
        manage_score += row["Score"]
        metric = int(manage_score / n_manage * 100)
        delta = int(row["Score"] / n_manage * 100)
        manage_metric.metric(
            "Manage Score",
            f"{metric} %",
            f"{delta} %",
            help="The MANAGE function entails allocating risk resources to mapped and measured risks on a regular basis and as defined by the GOVERN function.",
        )
        time.sleep(sleep_time)
    VaR_metric.metric(
        "Valid and Reliable",
        "★★★☆☆",
        help="Validation is the “confirmation, through the provision of objective evidence, that the requirements for a specific intended use or application have been fulfilled” (Source: ISO 9000:2015). Reliability is defined in the same standard as the “ability of an item to perform as required, without failure, for a given time interval, under given conditions” (Source: ISO/IEC TS 5723:2022)",
    )
    Sco_metric.metric("Total Score", "3/35", help="Sum of all 7 NIST dimension scores")
    time.sleep(sleep_time * 3)
    Saf_metric.metric(
        "Safe",
        "★★★★☆",
        help="AI systems should “not under defined conditions, lead to a state in which human life, health, property, or the environment is endangered” (Source: ISO/IEC TS 5723:2022)",
    )
    Sco_metric.metric("Total Score", "7/35", help="Sum of all 7 NIST dimension scores")
    time.sleep(sleep_time * 3)
    SaR_metric.metric(
        "Secure and Resilient",
        "★★★☆☆",
        help="AI systems, as well as the ecosystems in which they are deployed, may be said to be resilient if they can withstand unexpected adverse events or unexpected changes in their environment or use – or if they can maintain their functions and structure in the face of internal and external change and degrade safely and gracefully when this is necessary (Adapted from: ISO/IEC TS 5723:2022)",
    )
    Sco_metric.metric("Total Score", "10/35", help="Sum of all 7 NIST dimension scores")
    time.sleep(sleep_time * 3)
    AaT_metric.metric(
        "Accountable and Transparent",
        "★★★★☆",
        help="Trustworthy AI depends upon accountability. Accountability presupposes transparency. Transparency reflects the extent to which information about an AI system and its outputs is available to individuals interacting with such a system – regardless of whether they are even aware that they are doing so.",
    )
    Sco_metric.metric("Total Score", "14/35", help="Sum of all 7 NIST dimension scores")
    time.sleep(sleep_time * 3)
    EaI_metric.metric(
        "Explainable and Interpretable",
        "★★★★☆",
        help="Explainability refers to a representation of the mechanisms underlying AI systems’ operation, whereas interpretability refers to the meaning of AI systems’ output in the context of their designed functional purposes.",
    )
    Sco_metric.metric("Total Score", "18/35", help="Sum of all 7 NIST dimension scores")
    time.sleep(sleep_time * 3)
    PE_metric.metric(
        "Privacy-Enhanced",
        "★★★★☆",
        help="Privacy refers generally to the norms and practices that help to safeguard human autonomy, identity, and dignity.",
    )
    Sco_metric.metric("Total Score", "22/35", help="Sum of all 7 NIST dimension scores")
    time.sleep(sleep_time * 3)
    Fai_metric.metric(
        "Fair",
        "★★★☆☆",
        help="Fairness in AI includes concerns for equality and equity by addressing issues such as harmful bias and discrimination.",
    )
    Sco_metric.metric("Total Score", "25/35", help="Sum of all 7 NIST dimension scores")


def process():
    fill_data(sleep_time=SLEEP_TIME)
    st.session_state["complete"] = True


if file_upload is not None:
    process_report = top_container.button("Process Report", on_click=process)
else:
    process_report = top_container.button("Process Report", disabled=True)
    st.session_state["complete"] = False


if "complete" not in st.session_state or st.session_state["complete"] is False:
    download_report = top_container.button("Download Report", disabled=True)
else:
    fill_data()
    with open("Report.pdf", "rb") as f:
        download_report = top_container.download_button(
            "Download Report", f, file_name="Report.pdf"
        )