Key Points

References

Key Concepts

AI Governance Challenges

Google Gemini Pro has built-in anti-white bias

https://www.theverge.com/2024/2/21/24079371/google-ai-gemini-generative-inaccurate-historical

OpenAI warns not to use ChatGPT for hiring analysis

AI Deep Fake Image Challenges & Solutions

https://www.yahoo.com/finance/news/one-tech-tip-spot-ai-052451444.html

Experts say it might even be dangerous to put the burden on ordinary people to become digital Sherlocks because it could give them a false sense of confidence as it becomes increasingly difficult, even for trained eyes, to spot deepfakes.

USING AI TO FIND THE FAKES - limited access to these automated tools

Another approach is to use AI to fight AI.

Microsoft has developed an authenticator tool that can analyze photos or videos to give a confidence score on whether it's been manipulated. Chipmaker Intel's FakeCatcher uses algorithms to analyze an image's pixels to determine if it's real or fake.

There are tools online that promise to sniff out fakes if you upload a file or paste a link to the suspicious material. But some, like Microsoft's authenticator, are only available to selected partners and not the public. That's because researchers don't want to tip off bad actors and give them a bigger edge in the deepfake arms race.

AI Governance Concepts

Trusted AI

TRUSTED (Targeted, Reliable, Understood, Secure, stress-Tested, Ethical and Durable)

TRUSTED AI Strategy: Bank of England - 2024 - James Benford

Models will need to be TRUSTED (Targeted, Reliable, Understood, Secure, stress-Tested, Ethical and Durable) to ensure their ethical, safe and effective application across the organisation.

AI governance is critical to ensuring that we use AI in a way that enables experimentation and innovation while providing mitigants for risks, including data, model and third-party technology risks. We have completed an initial review of our internal policies^footnote[26] and put in place interim AI governance arrangements for our experimentation phase. We are completing a broader review, including by learning from others’ experience, as we continue with our AI pilots, work towards our AI strategy, and get ready to scale. The appropriate governance and safeguards for AI and its use is an evolving field; we will keep abreast with the evolution of international and domestic regulation, legislation, and best practice, to ensure we remain at the forefront of safe and responsible adoption.

Based on our initial work on internal uses, we believe there are several dimensions that need to be satisfied for AI models to be TRUSTED to inform our decision making at scale and effectively underpin our work to maintain monetary and financial stability. Running across all of them is the critical importance of humans being in the loop, in all aspects of the use of AI – development, deployment, operation and use – and both supported and held accountable for their role.

T is for Targeted. AI work needs to be focused on a tightly defined use case

R, is for Reliable. We are focused on reliable AI systems

U is for Understood. models themselves and the data that underlies them and their decisions and results need to be clear and comprehensible

S is for Secure.

T is for stress-Tested.

E is for Ethical.

E is for Evolving. Solutions and strategies need to evolve effecively and efficiently in line with good governance

D is for Durable. creating durable AI systems that can be sustained. Our data foundations and Data and Analytics need to continue to evolve

ADGM > Automated Digital Governance Manager from SWT

using original STEAR architecture and concepts, implements automated digital governance services engineered for a specific domain, goals, policies, procedures, transactions

Setup the governance domain

Track the transaction behavior for compliance with the policies and parameters for the specific procedures

Escalate the exceptions to the appropriate registered exception handler

Analyze the transactions for trends, changes, patterns identifying areas of concern

Remediate the actors, transactions, policies, procedures, parameters or goals to meet specific governance objectives in the domain

see the 2016 STEAR example on managing telemarketing behavior of auto dealers against regulations and OEM policies for compliance

Compliance Governance - supports regulations or standards

Compliance with Regulations or other standards.

Regulatory compliance may often be implemented in systems as data rules in programs, smart contracts etc

An IRS regulation requiring banks to report deposit details of greater than $5000 as part of AML programs is regulatory governance.

Operations Governance - supports operations policy effectiveness

Operations governance measures the effectiveness of policies using appropriate metrics.

Operations governance may be more dynamic, real-time.

AI and ML may add useful insights.

Using AI to identify patterns of accounts that structure multiple transactions always below $5000 within a 24 hour period may generate a SAR to investigate if the account is used money laundering effectively

generative-ai-WEF_Presidio_Recommendations_on_Responsible_Generative_AI_2023

Key Concepts

01 Establish precise and shared terminology

02 Build public awareness of AI capabilities and their limitations

03 Focus on human values and preferences

04 Encourage alignment and participation

05 Uphold AI accountability with rigorous benchmarknig and use case-specific testing while exploring new metrics and standards

06 Employ diverse red teams

07 Adopt transparent release strategies

08 Enable user feedback

09 Embed model and system traceability

10 Ensure content traceability

11 Disclose non-human interaction

12 Build human-AI trust

14 Develop comprehensive, multi-level measurement frameworks

15 Adopt sandbox processes

16 Adapt to the evolving landscape of creativity and intellectual property

17 Incentivize public-private research coordination

18 Build a common registry of models, tools, benchmarks and best practices

19 Support responsible open innovation and knowledge sharing

20 Enhance international collaboration on AI standards

21 Establish a global AI governance initiative

22 Prioritize social progress in generative AI development and adoption

23 Drive AI literacy across society

24 Foster holistic thought approaches in AI-driven environments

25 Steer generative AI’s transformative impact

26 Incentivize innovation for social good

27 Address resource and infrastructure disparities

28 Promote generative AI expertise within governments

29 Increase equitable access to AI in developing countries

30 Preserve cultural heritage

Proposed framework for Generative AI Goverannce article - 2024

PROPOSED MODEL AI GOVERNANCE FRAMEWORK FOR GENERATIVE AI-verifyf-2024.pdf. link

file

Open

generative AI2 has reinforced some of the same AI risks (e.g. bias, misuse, lack of explainability), and introduced new ones (e.g. hallucination, copyright infringement, value alignment).

This Model AI Governance Framework for Generative AI therefore seeks to set forth a systematic and balanced approach to address generative AI concerns while continuing to facilitate innovation. It requires all key stakeholders, including policymakers, industry, the research community, and the broader public, to collectively do their part. There are nine dimensions which the Framework proposes to be looked at in totality, to foster a trusted ecosystem.

key factors for trusted ai ecosystem

accountablity
data vcrs, 7v for model development, tuning, quality, accuracy
trusted dev & deployment
incident reporting
testing and assurance
security
content provenance
safety and r&d alignment
ai for public good

Open

AI Alliance for AI Governance - Meta and IBM

Meta, IBM Working With Startup to Test AI Model Safety

ai-governance-basic-concepts-ibm-2023.pdf link

file

principles of AI governance

levels of AI governance

Steps to implement AI governance

Visual dashboard

Health score metrics

Automated monitoring

Performance alerts

Custom metrics

Audit trails

Open-source tools compatibility

Seamless integration How organizations are deploying AI governance By adhering to these practices, organizations can establish a robust AI governance framework that supports responsible AI development, deployment and management, ensuring that AI systems are compliant and aligned with ethical standards and organizational goals.

add>>

fit to use cases for AI governance

Responsible AI Program - Microsoft

https://www.microsoft.com/en-us/ai/responsible-ai?activetab=pivot1%3aprimaryr6

Responsible AI Program - Google

https://ai.google/responsibility/responsible-ai-practices/

Google Gemini Image program has built-in “woke” bias creating people images

• Google Gemini has "woke" bias programmed in for image generation of people ***

  • article 1.       article 2.      article 3

The latest version of Google's Gemini artificial intelligence (AI) will frequently produce images of Black, Native American and Asian people when prompted – but refuses to do the same for White people.

In a statement to Fox News Digital, Gemini Experiences Senior Director of Product Management Jack Krawczyk addressed the responses from the AI that had led social media users to voice concern.

"We're working to improve these kinds of depictions immediately," Krawczyk said. "Gemini's AI image generation does generate a wide range of people. And that's generally a good thing because people around the world use it. But it's missing the mark here."

Gemini, formerly known as Google Bard, is one of many multimodal large language models (LLMs) currently available to the public. As is the case with all LLMs, the human-like responses offered by these AIs can change from user to user. Based on contextual information, the language and tone of the prompter, and training data used to create the AI responses, each answer can be different even if the question is the same.

Was Google ISO-42001 compliant on Responsible AI with Gemini Pro in 2024?

ISO-42001 - Responsible AI Concepts

https://aws.amazon.com/blogs/machine-learning/iso-42001-a-new-foundational-global-standard-to-advance-responsible-ai/

PWC Framework for AIG Governance

https://www.pwc.com/us/en/tech-effect/ai-analytics/managing-generative-ai-risks.html

AI LLM ( Large Language Model ) Governance and Modguard

https://www.forbes.com/sites/vipinbharathan/2023/06/25/guardrails-for-ai-what-is-possible-today/?sh=22df24b3a0de

ai-llm-modguard-governance-forbes.com-Guardrails For AI What Is Possible Today.pdf. link

Independent POGS - Proof of Governance As A Service

key concept that AI Governance can’t be done completely internal to an organization or a service even if based on standards

Inpendence provides trust verifications needed on compliance with AI governance policies at runtime

Sybal example of Dynamic POGS in banking

capital markets, securities management concepts | capitalmarkets,securitiesmanagementconcepts SybalPOGSexplainer

Gen AI content harvesting governance for LLMs

Nightshade and Glaze provide tools to poision LLMs on harvested artistic works

https://www.technologyreview.com/2023/10/23/1082189/data-poisoning-artists-fight-generative-ai/

Tracking harvesting > embed traceability for origins / provenance in images etc and track harvested access from a managed content site ( not normal sites )

AI Governance Concepts

SGS delivers STEAR governance capabilties - Jim Mason

https://skywebteam.atlassian.net/wiki/spaces/KHUB/pages/426836356/m+Solution+Ideas+1#mSolutionIdeas1-ComplianceTracker-STEAR

How SGS provides STEAR governance capabilities >>

SGS ( Smart Governance Services support BZT ( Better Zero Trust ) with self governance and automated health checks )

The STEAR governance life cycle supports a variety of governance use cases for AI and any other process - dynamic or batch processing

S = Setup principles, policies, processes for governance

T = Track behavior and transactions with pre-execution verifications, execution monitoring, post-execution analytics

E = Escalation of events automatically from the process execution to registered listeners and handlers 

A = Audit automation 

R = Remediation automation based on policies, processes and conditions

Unleashing the power of AI for Humanity

AI Regulations

EU-AI-ACT-RULES ON ARTIFICIAL INTELLIGENCE-2024

https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52021PC0206

Chintan Oza

https://www.linkedin.com/posts/chintanoza_the-state-of-ai-regulations-2024-activity-7147951909058908160-X0zg?utm_source=share&utm_medium=member_desktop

• 2nd• 2ndEnabling Ecosystem for Startups & Investors in G20 & BRICS | Your Sherpa in Digital & beyond | Author #RamAayenge SeriesEnabling Ecosystem for Startups & Investors in G20 & BRICS | Your Sherpa in Digital & beyond | Author #RamAayenge Series

The AI market is booming! It's set to grow by 19.6% annually and reach $500 billion this year.

With the adoption of AI across industries, regulatory focus has also increased at a similar pace.

A recent publication by Holistic AI takes a closer look at AI regulations in the EU, US, China, and UK, providing a high-level breakdown of what to expect in 2024.

Stay informed and stay ahead of the game!

ai-gov-The State of Global AI Regulations in 2024.pdf. link

file

Looking at the EU, US, China, and UK this publication sheds light on current approaches to AI regulation around the world, while providing high-level breakdowns of what to watch for in 2024.

US AI Regulation and Bills

Fault lines emerge on AI Senate report - 2024

US-AI-senate-bill-issues-2024-Fault lines emerge on AI Senate report link

Is there any value in co-opting AI regulations for everyone with AI regulations based on identity political groups???

Biden team thinks so

Candidate Solutions

OpenLLama - Open LLM

https://skywebteam.atlassian.net/wiki/spaces/KHUB/pages/2540732417/OpenLlama+LLM

Securiti - steps to responsible AI

https://links.securiti.ai/steps-to-responsible-ai/

Anthropic - an AI safety and research company working to build reliable, interpretable, and steerable AI systems.

https://www.linkedin.com/company/anthropicresearch/

We're an AI research company that builds reliable, interpretable, and steerable AI systems. Our first product is Claude, an AI assistant for tasks at any scale. Our research interests span multiple areas including natural language, human feedback, scaling laws, reinforcement learning, code generation, and interpretability.

https://www.anthropic.com/

https://www.anthropic.com/product

Report on Future AI Generative Tools - 2023

ai-tools-future-generative-ai-tools-report-2023 link

ai-tools-future-generative-ai-tools-report-2023 file

Some AI apps, tools and plugins

Chatgpt Chrome extensions add value for users

https://cointelegraph.com/news/5-chatgpt-chrome-extensions

WebChatGPT

The useful plugin WebChatGPT improves an individual’s ChatGPT prompts by including pertinent web search results. Users will benefit from more precise and relevant web searches using this extension.

The ability to retrieve web results for one’s queries, extract text from any URL of a webpage, and add and use prompt templates are some of WebChatGPT’s primary capabilities. With the help of these capabilities, one can communicate with ChatGPT while taking advantage of the huge online resources.

ChatGPT for Google

ChatGPT for Google is a browser extension that enhances search engines by integrating the capabilities of ChatGPT. The responses from ChatGPT can be seen alongside standard search engine results when using this add-on.

Currently, search engines such as Google, Baidu, Bing, DuckDuckGo, Brave, Yahoo, Naver, Yandex, Kagi and Searx are supported by ChatGPT for Google. One needs a ChatGPT account to use this extension, which can be created at chat.openai.com for free.

Google Deep Mind

DeepMind Says Its AI Solved a Math Problem That Humans Were Stumped By

https://www.yahoo.com/news/deepmind-says-ai-solved-math-123036185.html

DeepMind claims that for the first time, an AI has solved a famously difficult math problem with a solution that eluded human mathematicians — which could be huge if it holds up to scrutiny.

Feeding code into an AI is one thing, but having it spit out a brand-new solution to a famous puzzle — even though it took a few days, as MIT Tech points out — is a different thing entirely.

"It’s not in the training data," DeepMind research VP Pushmeet Kohli told the website. "It wasn’t even known."

There is something of a mystical quality to what the DeepMind scientists are claiming: that the LLM managed to — just maybe — think for itself.

Google BARD

<<#BUG>> new Confluence page editor won’t allow links as text any more

must navigate to the page then copy the url from the page - not a link

https://bard.google.com/?utm_source=sem&utm_medium=paid-media&utm_campaign=q4enUS_sem7

Google Bard - Getting Started Guide

x Grok

https://www.yahoo.com/finance/news/elon-musk-debuts-rebellious-grok-015320778.html

Compose AI

Compose AI is a handy Chrome extension that improves users’ writing by providing auto-complete ideas and speeding up typing. It is a perpetually free plugin, but there is also a commercial version that offers more sophisticated features, including tailored recommendations.

One may easily produce emails, Slack conversations or large essays using Compose AI. As users type, the plugin dynamically adjusts to the various writing contexts and offers them the most appropriate and pertinent ideas. One may increase the quality of their written communication and save time by utilizing Compose AI.

TeamSmart AI

TeamSmart AI offers a number of capabilities that enrich the ChatGPT experience. Among other features, users can check the availability of a domain, customize their code theme, ask questions regarding the material on the current tab, and use markdown for formatting. It’s crucial to remember that all data is securely and locally kept, guaranteeing the privacy of users’ data and preventing any sharing with outside parties.

WiseOne

Wiseone is a sophisticated browser add-on created to improve the experience of reading and researching online. This cutting-edge addition, which is powered by AI technology, enables users to easily get accurate information, master a variety of subjects and simplify complex material.

Wiseone provides a wide range of features to improve the reading experience, including:

• Discover: Users can delve into articles and grasp intricate concepts and unfamiliar terms, enabling them to fully comprehend the content they’re reading.

• Cross-checking: Wiseone enables fact verification by comparing information from multiple sources on the same subject, ensuring accuracy and reliability.

• Ask anything: Leveraging OpenAI’s ChatGPT, users can obtain simplified and understandable answers to complex information, making comprehension effortless.

• Summarize: With the summarization feature, users can efficiently read through content while retaining key information, facilitating effective knowledge retention.

• Suggestions: Wiseone provides curated suggestions of articles from diverse sources, allowing users to deepen their understanding of a particular subject.

Prompt Perfect

https://xyz-prompt-perfect.uc.r.appspot.com/#

Prompt Perfect optimizes your prompts in ChatGPT to ensure high-quality, relevant responses.

Begin your prompt with 'perfect' and watch as your input is automatically refined to perfection

Wolfram

Wolfram for Math and more precise responses

https://www.youtube.com/watch?v=AsS5lIsirVA

If you use ChatGPT for a lot of coding or mathematics, then Wolfram is a must-have ChatGPT productivity plugin. This plugin essentially takes the effectiveness of ChatGPT when it comes to mathematics and increases it dramatically. If you’ve seen the viral tweets about people using ChatGPT to calculate a weight loss regimen and want to do the same yourself, this is the plugin you need.

Top 10 ChatGPT plugins

https://www.youtube.com/watch?v=o2M_paJf48I

More ChatGPT plugins for Chrome Browser

https://www.searchenginejournal.com/chatgpt-chrome-extensions/485594/

ChatGPT for Google

showcases ChatGPT’s replies right next to search results from popular engines like Google, Bing, and DuckDuckGo.

You can chat with ChatGPT about the query and receive answers.

It also has features like a dark mode, markdown rendering, code highlights, and custom trigger mode. As a bonus, it supports ChatGPT Plus and GPT-4 (manual enable needed) and offers a way to give feedback for ChatGPT improvements.

Usage policy

https://searchgpt.net/privacy-policy.html

WebChatGPT

The biggest limitation of ChatGPT is that its knowledge base is limited to 2021.

With WebChatGPT, however, this is no longer an issue. This Chrome extension adds relevant web results to the prompts to ensure the conversations are more accurate and up-to-date.

It lets you get web results for your queries and extract webpage text from any URL. The additional prompt templates and use of DuckDuckGo bangs are super helpful in getting more precise search results from thousands of websites.

ChatGPT Writer for emails and messages only

With ChatGPT Writer, you can tap into ChatGPT’s capabilities to craft emails and messages on any website.

ChatGPT for Chrome

enhances your search experience by displaying ChatGPT’s responses alongside search engine results from Google, Bing, and other search engines.

This free and convenient extension lets you access ChatGPT’s advanced AI capabilities directly from your Chrome browser.

Its key features include code highlights for easy understanding, a dark mode for comfortable viewing, and a clipboard copy option for pasting information elsewhere.

Merlin - clicking Cmd+M, you can access ChatGPT on any website - summarize youtube videos

is an amazing ChatGPT extension that works across your entire browser. By clicking Cmd+M, you can access ChatGPT on any website – from Google to Gmail and beyond.

It’s free and privacy-first, and it helps you save time with features like summarizing YouTube videos, generating email responses, and creating AI-driven social engagement on platforms like Twitter and LinkedIn.

ChatGPT Prompt Genius

is a free and open-source browser extension designed to provide you with an extensive library of prompts, customization options, and additional features.

You can easily discover, share, import, and use the best prompts for ChatGPT while syncing your chat history locally for easy access and searching.

You can export your chats in various formats, such as markdown, HTML, PDF, or PNG

Engage AI - writes LInkedin comments

is a ChatGPT Chrome extension specifically designed to help you write insightful LinkedIn comments, build relationships, and establish connections with prospects. It lets you generate memorable, authentic, insightful, and relevant comments.

Comments need review before publishing

Summary with ChatGPT and Claude is a versatile Chrome extension that uses OpenAI’s AI technology to offer clear and concise summaries of YouTube videos and articles.

It has various features, such as generating bullet point summaries for articles, customization of ChatGPT prompts, or support for both OpenAI API Key and ChatGPT login.

It also offers example prompts, such as summarizing an article in 3 bullet points or 50 words, translating text to Spanish, simplifying text for a 5-year-old, and writing a caption based on the content.

How to Summarise YouTube Videos on ChatGPT (How to Summarize a YouTube Video with Chat GPT)

https://www.youtube.com/watch?v=QrHizi-CSME

https://chromewebstore.google.com/detail/youtube-summary-with-chat/nmmicjeknamkfloonkhhcjmomieiodli?utm_source=ext_app_menu

Talk-to-ChatGPT

is perfect for when you don’t want to type your queries. It allows you to talk to ChatGPT using your microphone, converts speech to text, and delivers AI-generated responses audibly.

The official extension supports all major languages and is a breeze to use on the ChatGPT homepage.

It’s a fantastic way to simplify your ChatGPT experience, as it utilizes Google Chrome’s speech recognition and text-to-speech technologies.

ReaderGPT

can be a lifesaver for managing your online reading. It generates summaries of any text-based webpage using ChatGPT to help you quickly get the gist of long articles or blog posts.

Its vast array of example prompts in the settings allows you to find the best way to summarize the content you are interested in.

AWS Generative AI Example - Image to Speech

https://www.linkedin.com/posts/grazia-russo-lassner-2165894b_aws-activity-7072461280321691648-wwPA?utm_source=share&utm_medium=member_desktop

A great #generativeAI example of how to orchestrate multiple #AWS services for an original image-to-speech solution.

https://aws.amazon.com/blogs/machine-learning/introducing-an-image-to-speech-generative-ai-application-using-amazon-sagemaker-and-hugging-face/

Amazon is touting the fact that Q was trained on 17 years of AWS data as a key differentiator for the platform, which is currently available in preview. In one example, Amazon explains how you can use Q to take information from a corporate blog and turn it into a social media post complete with hashtags. In another example, the company explains how you can use Q to create helpful apps for workers.

The service is similar to what Microsoft and Google currently offer across their generative AI platforms. Microsoft’s Copilot and Azure AI studio help workers summarize content and craft emails, as well as create custom AI copilots. Copilots are essentially AI-powered apps.

Google AI pricing

Google’s Duet for Workspaces, meanwhile, acts as an generative AI helper for brainstorming and summarizing content, while its Vertex AI services helps enterprise customers build generative AI apps

Duet pricing = Microsoft 365 AI pricing = $30 per user per month

Vertex AI pricing

https://cloud.google.com/vertex-ai/pricing

Step-by-step guide for Example

sample code block

sample code block

AI Tools for Marketing

AI Marketing Training and Community for the Modern Marketer

free AI tools courses

Recommended Next Steps

Related articles