Top GenAI Products By Visitors

 

Gartner GenAI Framework

Data Structures

bytebytego

Software Architecture

bytebytego

Graph Rag

Hallucination Leaderboard

vectara hallucination leaderboard

Important Events of 2023

Events of 2023

EU AI Act

EU AI Act

USA Failed Wars

• Vietnam War
• Bays of Pigs Invasion
• Korean War
• Russian Civil War
• Second Samoan War
• Formosa Expedition
• Red Cloud's War
• Powder River Indian War
• War of 1812
• Afghanistan War
• Iraq War

Falcon LLM

Falcon LLM

Ethical Consumer

Ethical Consumer

21st Century Biggest Data Breaches

• Yahoo
• Aadhaar
• Alibaba
• Linkedin
• SinaWeibo
• Facebook (Meta)
• Marriott International (Starwood)
• Adult Friend Finder
• MySpace
• NetEase
• Course Ventures (Experian)
• Dubsmash
• Adobe
• Capital One
• Target
• Heartland Payment Systems
• Equifax
• eBay
• Hold Security
• TJX Companies
• JPMorgan Chase
• US Office of Personnel Management
• Sony's PlayStation Network
• Anthem
• RSA Security
• Stuxnet
• Verisign
• HomeDepot
• Uber
• Microsoft
• First American Financial
• Cambridge Analytica
• RiverCity Media
• Exactis
• DeepRoot
• Zynga
• Plex Movies & TV
• LAUSD Unified
• Cash App
• CAM4
• Verifications IO
• Twitter
• Sociallarks
• Deep Root Analytics
• MyFitnessPal
• Canva
• Apollo
• Badoo
• Evite
• Quora
• VK
• MyHeritage
• Youku
• Rambler
• Dailymotion
• Dropbox
• Tumblr
• Ashley Madison
• LastPass
• Bonobos
• MGM Grand
• Optus
• Medibank
• Easyjet
• 123RF
• Twitch
• Neiman Marcus
• MeetMindful
• Pixlr
• Tackle Warehouse, Running Warehouse, Tennis Warehouse, SkateWarehouse
• Harbour Plaza Hotel Management
• Graff
• Zoom
• Slickwraps
• Magellan Health
• Nintendo
• Mailfire
• Solarwinds
• Pegasus Airline
• Philippines Comelec
• MailChimp

Generate JavaScript Model

• Alpine.js
• Stimulus.js

Cloud Gaming Option

Cloud Gaming Options

Thirdparty Forms

• Google Forms
• WuFoo
• Netlify Forms
• Formspree
• FormKeep
• Getform

form providers

Icon Fonts

• FontAwesome
• Lineicons
• TheNounProject
• Material Design Icons
• FlatIcon
• IconFinder
• Feather Icons
• StreamLineIcons
• Linearicons
• Unicons
• NucleoApp
• CoreUI Icons
• Line Awesome
• StackIcons
• Twemoji Awesome
• Font Diao
• Sociolicious
• Octicons
• Android Icons
• DevIcons
• Open Iconic
• Fontello
• Boxicons
• Ionicons
• Boostrap Icon
• Google Icons
• Glyphicons
• Typicons

Recruitment Processes

Making job applications can be frustrating. No one has the time to tailor their resume to each role. And, what is worse is that the recruiter is likely to only be hunting for keywords. If they use an application tracking system it will likely have its own approach to matching. The hiring manager will have their own needs. Most companies have a failed recruitment process where many able candidates are rejected through inadequate recruiters and systems. Automation does not help candidates. However, everyone needs to be given a fair chance of review. Unprofessionalism is typical in most recruitment processes. What is worse is when a candidate goes through multiple stages and towards the end the employer is unable to get funding approval for new hires. Not only is it unfair and frustrating for candidates but it also wastes time on both sides. Streamlining recruitment processes is important for every responsible organization as it is the first point of contact for individuals. Bad recruitment processes is a negative reflection of an organizational practices and the way they treat customers. If the individual has a bad experience as a candidate, they will likely be less inclined to join the company as an employee. All in all, bad recruitment processes can affect the reputation of an organization. 

Coding Tests

Organizations use third-party coding platforms for testing candidates as part of recruitment process.  Examples of such platforms include: HackerRank, Leetcode, Codility, Qualified, among others.  In many respects, use of such platforms in recruitment practice is counterproductive for several reasons, as stated below. 

• The problem statement is usually unclear and not defined in similar terms that would be practical for a business case
• Often the problem is defined in mathematical terms which is likely to confuse the candidate
• The testing environment is often buggy
• Tests are very localized and mundane
• Tests can drive bad practices
• Tests can reflect problems that candidate will never have to solve in practice
• Tests have incorrect test cases
• Tests are erratic
• Tests don't lend themselves very well to practical business use cases
• Tests ignore disabilities, social adjustments, and are counter to diversity, inclusion, and equity
• Tests have tedious edge cases and ambiguous logic
• Tests develop a sense of social distrust especially with experienced candidates
• Tests have questions that are not tailored to skills required for the job
• Candidates can pass even if they lack the necessary skills for the job
• Candidates who have a lot of practice on the platform can game the system
• Harder to code in an unfamiliar environment
• Tests can filter out good candidates leaving mediocre and average candidates
• Tests tend to be based on algorithmic skill rather than the ability to code
• Tests require time which can be defined as chargeable time especially if it requires developing a model
• Tests are often bookish in nature and rarely a reflection on a candidate's experience
• Tests can reset haphazardly in middle of a session
• Tests tend to be timed meaning they don't provide sufficient flexibility to candidates to fit into their busy schedules
• Tests can be done using Generative AI

Graph RAG

RAG

RAG

Graph RAG

Graph RAG Query

Jamstack Themes

Jamstack Themes

Jamstack Ecosystem

Jamstack Ecosystem

Bejamas Discovery

Onprem LLM

OnPrem LLM

PrivateGPT

Prompt Engineering Guide

Prompt Engineering Guide

Content Dubbing

In many cases, subtitles on dubbed content does not match. Often the dubbed speech does not match the actor's original speech. And, many times the nuances and differences in languages don't match up. Even the dubbed speech and subtext is out of sync with the image frames. What would really help here are few things:

• Quality subtext to match original spoken language
• Actors original voice matched and morphed to the translated dubbed language
• Sync the translations with the image key frames
• This will require video and speech recognition models that have quality training datasets

Should humans trust ChatGPT?

Trust in ChatGPT, or any AI model, should be approached with caution and consideration of its limitations. Here are some key points to keep in mind regarding trust in ChatGPT:

1. Limited Understanding: ChatGPT, like other AI models, lacks true understanding or consciousness. It generates responses based on patterns and data it has seen during training. It may not always provide accurate or contextually appropriate answers, especially in complex or nuanced situations.

2. Bias and Fairness: AI models like ChatGPT can inherit biases from their training data. Trusting it blindly without critical evaluation can lead to biased or unfair responses. It's important to be aware of potential biases and take steps to mitigate them.

3. Transparency: AI models like ChatGPT operate as "black boxes," making it challenging to understand how they arrive at specific responses. This lack of transparency can make it difficult to trust the reasoning behind their answers.

4. Responsibility: Trust in ChatGPT should be accompanied by a sense of responsibility. Users and developers have a responsibility to verify information provided by the AI and to use it ethically.

5. Verification: It's advisable to verify information obtained from ChatGPT, especially when making critical decisions based on its responses. Don't rely solely on the AI for matters that require accuracy and reliability.

6. Content Moderation: Implement content moderation when using ChatGPT in public-facing applications to prevent the generation of inappropriate or harmful content.

7. User Education: Users should be educated about the capabilities and limitations of ChatGPT. Understanding its nature as a machine learning model can help users make informed decisions about the information they receive.

8. Ethical Use: Trust should be based on ethical use. Developers and organizations should use ChatGPT responsibly, following ethical guidelines and best practices.

In summary, while ChatGPT and similar AI models can be valuable tools for generating content and providing information, trust should be tempered with a critical and responsible approach. It's essential to be aware of their limitations, potential biases, and the need for verification, especially in situations where trust is critical. Trust should be earned through responsible usage and ongoing efforts to improve model performance and fairness.

How ChatGPT formulates an answer to a question

ChatGPT, based on the GPT-3 architecture, formulates an answer to a given question. ChatGPT uses a combination of pre-trained knowledge and pattern recognition to generate responses. Here's a simplified breakdown of its process:

1. Pre-training: ChatGPT is pre-trained on a vast corpus of text from the internet. During this phase, the model learns to predict the next word in a sentence based on the context of the words that came before it. This process helps the model understand grammar, syntax, and semantic relationships between words.

2. Context Understanding: When you input a question or statement, ChatGPT uses the context provided to understand what you're asking. It analyzes the words, their order, and the relationships between them.

3. Pattern Recognition: ChatGPT has learned patterns and associations from its pre-training data. It identifies patterns in the input text that match patterns it has seen during training. This allows it to recognize similar questions or topics and generate relevant responses.

4. Probability Estimation: The model assigns probabilities to each possible word or token that could come next in the response. It estimates these probabilities based on its understanding of the context and the patterns it has learned.

5. Generation: ChatGPT generates a response by selecting the word or token with the highest probability as the next word. It repeats this process to generate a sequence of words that form a coherent response.

6. Contextual Awareness: Importantly, ChatGPT is aware of the context of the conversation. It remembers the conversation history, so it can refer back to previous messages to maintain context and coherence in the conversation.

7. Fine-tuning: In some applications, ChatGPT can be fine-tuned on specific tasks or domains. Fine-tuning helps adapt the model to perform better in specific contexts, such as customer support or content generation.

It's important to note that while ChatGPT can generate impressive responses, it doesn't possess true understanding or consciousness. It operates based on patterns and probabilities learned from text data. Its responses are generated based on statistical associations, and it may not always provide accurate or contextually appropriate answers.

Additionally, the quality of responses can vary, and ChatGPT may generate biased or inappropriate content, which highlights the need for responsible usage and content moderation when implementing such models.

Is ChatGPT PII Compliant?

As of last update in September 2021, ChatGPT and similar AI models should be used with caution when it comes to handling personally identifiable information (PII). These models do not have built-in PII compliance mechanisms, and using them for tasks that involve PII without proper safeguards can raise privacy and compliance concerns. Here are some important considerations:

1. Data Handling: When interacting with ChatGPT, you should avoid sharing or inputting sensitive PII, such as Social Security numbers, credit card information, addresses, or other personal identifiers. If you must handle PII, it's essential to do so in a secure and compliant manner.

2. Data Encryption: If you need to exchange PII with an AI system, ensure that data transmission and storage are encrypted and follow industry-standard security protocols to protect sensitive information.

3. User Consent: Obtain informed consent from users before collecting or processing their PII. Clearly communicate how the data will be used and stored.

4. Data Retention: Minimize the retention of PII and only store it for as long as necessary. Implement data retention policies that comply with relevant data protection regulations.

5. Compliance with Regulations: Ensure that your use of ChatGPT complies with data protection regulations such as the General Data Protection Regulation (GDPR) in Europe, the Health Insurance Portability and Accountability Act (HIPAA) in the United States, or other relevant laws in your jurisdiction.

6. Privacy Impact Assessment: Conduct a privacy impact assessment to evaluate the potential risks and privacy implications of using ChatGPT with PII. Implement necessary safeguards to mitigate these risks.

7. User Anonymization: Whenever possible, use anonymized or de-identified data for training and testing AI models to minimize privacy risks.

8. Ethical Use: Adhere to ethical principles in AI development and usage, including transparency, fairness, and accountability, especially when dealing with PII.

It's important to note that PII compliance is a complex legal and ethical matter that may vary depending on your location and the specific context of use. Compliance often involves legal and data protection experts to ensure that your AI system, including ChatGPT, complies with relevant regulations and safeguards user privacy.

Since the landscape of AI and compliance is continuously evolving, it's advisable to consult with legal experts and stay informed about any updates or changes in data protection laws and regulations that may affect your use of ChatGPT or similar AI technologies.

Is ChatGPT Ethical?

The ethics of using ChatGPT or similar AI models depend on how they are used and the context in which they are applied. Here are some key ethical considerations:

1. Bias and Fairness: AI models like ChatGPT can inherit biases from the training data, which may result in biased or unfair responses. It is crucial to continually monitor and mitigate bias to ensure fair and unbiased interactions.

2. Misinformation: ChatGPT can generate text based on the input it receives, including potentially inaccurate information. Users and developers have a responsibility to verify and fact-check information obtained from AI models.

3. Privacy: Using AI models for sensitive data or conversations can raise privacy concerns. Care must be taken to protect personal information and adhere to privacy regulations.

4. Malicious Use: AI models can be used for malicious purposes, such as generating spam, deepfakes, or engaging in harmful online activities. Developers should implement safeguards to prevent misuse.

5. Transparency: Users should be aware that they are interacting with AI, and transparency about the AI's capabilities and limitations is essential.

6. Informed Consent: In certain contexts, it may be necessary to obtain informed consent from users if AI is involved in sensitive or personal interactions.

7. Job Displacement: The automation of tasks by AI can impact employment in some industries. Ethical considerations should include strategies for workforce adaptation and job displacement mitigation.

8. Environmental Impact: Training and running large AI models like ChatGPT require significant computational resources and can have an environmental impact. Minimizing this impact is an ethical concern.

9. Monitoring and Oversight: Continuous monitoring and oversight of AI systems are crucial to identify and address ethical issues as they arise.

It's important to note that the responsibility for the ethical use of AI models like ChatGPT lies with developers, organizations, and users. Many organizations and researchers are actively working on guidelines and best practices to ensure the ethical deployment of AI technologies. Ethical AI development includes not only creating responsible models but also addressing societal and cultural concerns associated with their use.

Ultimately, whether ChatGPT is ethical or not depends on how it is implemented and used, as well as the ethical principles and guidelines followed by the parties involved.

Transformers, LLMs, and Embeddings

"Transformers," "Large Language Models (LLM)," and "Embeddings" are all related concepts in the field of natural language processing (NLP) and deep learning, but they refer to different aspects of NLP models and techniques. Here's a breakdown of the differences:

1. Transformer Models:

   • Definition: The Transformer is a deep learning architecture introduced in the paper "Attention Is All You Need" by Vaswani et al. in 2017. It was designed to handle sequential data efficiently, making it particularly well-suited for NLP tasks.
   • Key Features: The Transformer architecture relies heavily on self-attention mechanisms to process input sequences. It consists of an encoder-decoder structure, but for tasks like language modeling and text generation, only the encoder is often used.
   • Applications: Transformers are used as the underlying architecture for various NLP models, including large language models like GPT-3 and BERT. They have proven highly effective for tasks such as text classification, translation, summarization, and more.

2. Large Language Models (LLMs):

   • Definition: Large Language Models are a specific type of model built on the Transformer architecture, designed to understand and generate human-like text. They are characterized by their enormous size, often containing tens of millions to hundreds of billions of parameters.
   • Key Features: LLMs are pre-trained on vast amounts of text data from the internet, allowing them to learn language patterns, facts, and even some reasoning abilities. They can be fine-tuned for specific NLP tasks.
   • Applications: LLMs can be used for a wide range of NLP tasks, such as text generation, translation, sentiment analysis, chatbots, and more. GPT-3 and BERT are examples of LLMs.

3. Embeddings:

   • Definition: Embeddings are representations of words or tokens in a continuous vector space. They are a fundamental component of NLP models and are used to convert discrete words or tokens into numerical vectors that can be processed by neural networks.
   • Key Features: Word embeddings, such as Word2Vec, GloVe, and FastText, map words to vectors based on semantic and syntactic relationships in the training data. These embeddings capture contextual information and are used as input features for NLP models.
   • Applications: Embeddings are used in a wide variety of NLP tasks, including word similarity calculations, text classification, sentiment analysis, and more. They enable models to work with words as numerical data, facilitating the learning of complex language patterns.

In summary, Transformer models are a type of architecture used for sequence processing, with the Transformer architecture being the foundation. Large Language Models (LLMs) are a specific application of Transformer models designed for understanding and generating human-like text, characterized by their size and pre-training on vast text corpora. Embeddings, on the other hand, are representations of words or tokens in vector space, used as input features for NLP models to enable them to process text data effectively. LLMs often use embeddings as part of their architecture for token-level representations.

What are Large Language Models?

Large Language Models (LLMs) are advanced artificial intelligence models designed to understand and generate human-like text. They belong to a broader category of machine learning models known as natural language processing (NLP) models. LLMs have become increasingly prominent in recent years due to their ability to perform various language-related tasks, such as text generation, translation, sentiment analysis, question answering, and more.

Here are some key characteristics and details about large language models:

1. Size: LLMs are characterized by their enormous size in terms of parameters, often ranging from tens of millions to hundreds of billions of parameters. A parameter in this context refers to a tunable aspect of the model that it learns from training data.

2. Pre-training and Fine-tuning: LLMs are typically pre-trained on vast amounts of text data from the internet. During pre-training, they learn language patterns, grammar, facts, and even some reasoning abilities. After pre-training, they can be fine-tuned on specific tasks or domains, which helps adapt them for various applications.

3. Transformer Architecture: Most LLMs, including GPT-3 and BERT, are built on the Transformer architecture. Transformers have revolutionized NLP by allowing models to capture long-range dependencies in text, making them highly effective for a wide range of language tasks.

4. Versatility: LLMs are versatile and can be used for a variety of NLP tasks. They can be fine-tuned for specific applications such as chatbots, language translation, text summarization, and more. This versatility has made them valuable in industries like healthcare, customer service, content generation, and beyond.

5. Ethical and Bias Considerations: LLMs have raised ethical concerns related to bias and misinformation. Since they learn from the vast and sometimes biased internet data, they can inadvertently reproduce biases present in the data. Efforts are ongoing to mitigate these issues and make LLMs more responsible.

6. Computationally Intensive: Training and using LLMs require significant computational resources, including powerful GPUs or TPUs. This has limited access to such models to organizations with substantial computing resources.

Examples of popular large language models include GPT-3 (Generative Pre-trained Transformer 3), BERT (Bidirectional Encoder Representations from Transformers), and T5 (Text-to-Text Transfer Transformer).

LLMs have had a profound impact on various industries, from improving language translation services to enabling more advanced chatbots and content generation tools. They continue to advance and have the potential to reshape the way humans interact with computers and information.

Unsplash

Unsplash

Static Site Generators

Static Site Generators

Llama 2

Llama 2

LLMOps

• LLMFlow
• LangChain
• Haystack
• LlamaIndex
• PromptTools
• EmbedChain
• Dify
• DeepLake
• BudgetML
• GPTCache
• LangFlow
• xTuring

LLMOps

SeamlessM4T

SeamlessM4T

Grant Funding Assessors

Assessors that evaluate grant funding proposals are on the whole quite clueless. In fact, their understanding of the project for grant proposal is rather low. They make generally terrible assessors with their background and experience. And, often their biases play a big part in how they quantify scores to a grant proposal. The feedback is usually a pretty good indicator of how utterly clueless, out of place, and unqualified they are in their designated roles. There are plenty of examples of projects that have failed to secure grant funding because of biased assessor. Many have also started using ChatGPT and other autogenerated tools to produce feedback summaries that may even be out of context through model hallucination. The entire grant funding process needs to be automated so the human-in-loop factor is removed with an objective peer-review process. Often assessors are associated to an organization who will likely be sponsored by other organizations. This on its own can lead to a conflict-of-interest and a biased outcome of applications that are filtered out through the already fairly subjective process of evaluations. The other aspect is that the various sections also require repeating the same information over and over again which is rather useless for all intents and purposes. On the whole, assessors have a very biased way of assessing applications.

Teiid

Teiid

ArcadeDB

ArcadeDB

Pyodide

Pyodide

Application Tracking Systems

Application Tracking Systems are the bane of organizational recruitment. They reject 75% of applications while accepting only 25%. And, majority of this is down to formatting. Even if you are lucky enough to get past to the recruiter, they are also likely to be interested in counting keywords. In most organizations recruitment processes are antiquated and handled by people that don't even understand the hiring needs. Furthermore, machine learning models compound the issue by automating the process of counting word distributions which is what was done manually by the recruiter. In essence, not really solving the issue of understanding semantics and context towards looking beyond the surface of a document and into discourse processing. The following things are typical with ATS:

• Submit an application and receive a standard rejection email response within minutes. There is no way that a recruiter could have reviewed the application that quickly.
• Because the automated systems are also looking at word distributions they miss out on the context and content of work.
• It is very common to get rejected by an ATS but get approached by a talent partner for the very same role.
• Since it is looking at word distributions, having two many keywords confuses the automated system
• It seems to also look at job titles, what if the job title did not exist in the market for the length of experience required.
• What if the technical skill required did not exist for the length of experience required, asking for ten years of experience in LLM will get a lot of CVs rejected because they are very unlikely to have so many years of experience. Or, the fact that you mention tons of experience with embedding models but no mention of LLMs. 
• Acronyms and abbreviations are tricky.
• Skills are not properly evaluated in context.
• Formating of CV seems to be the end game. But, if that is the case then it defeats the purpose because after passing the ATS it will be screened by the recruiter and then the manager who likely has their own requirements.
• If the system rejects 75% of CVs then it's proof that the system is not working.
• If the talent partner then later also approaches after the ATS rejection then it is further proof that the system does not work.
• If the ATS + the recruiter rejects you for the role that you are a close match compared to the candidates that have very little experience in the area, then is very likely that there is serious bias in the recruitment process.
• At times, the entire recruitment process is done to legally filter out minorities and further hides the curtain of institutional racism within an organization.
• It may even ask for information about protected characteristics hiding under the covers of initiatives for diversity, equity, and inclusion to transform them into discriminatory filters.
  
• Hiring for keywords and job titles is a useless way to match jobs to candidate profiles, or better yet to build a candidate profile in first place.
• No one has time to customize their CV to every job to game the ATS.
• ATS resume parsing is useless, if it is based on keywords, job titles, file type, formating, and special characters.
• ATS ignores aspects of discourse processing which is the essence of resume parsing.
• If you apply to more than one role the ATS might consider that as spam.
• It also proves that human resources is the weakest link when an organization states that there is a skills shortage in the market when there really isn't. The issue is with rejecting 75% of applications at the expense of using an ATS process, some of which are perfectly good candidate profiles to the job requirements.
• Rejecting a CV just because it doesn't meet a particular formatting criteria is not really a reason to decline a candidate application for a job.
• This is a case for where AI Automation is bad, likely unethical, untrustworthy, lacks assurance, and irresponsible.
• Recruiters are terrible at reading and understanding CVs, and ATS makes it even worse
  

Survey of Large Language Models

Survey of Large Language Models

M0deler

M0deler

Paper QA

Paper QA

Argumentation Structures

 

Taxonomy of Research Publications

 

Taxonomy of Transformers

Evolutionary Tree

LLM App Stack

AGI

For Artificial General Intelligence to see reality, there has to be an extension in the use of Large Language Models that comprise of short-term memory, long-term memory, and sensory memory to provide for an abstraction in associative memory in implicit and explicit forms. This will also need to extend into some form of representation in cognitive modeling as well as a quantum information to extend into space and time geometry. And, above all an aspects of sapience, self-awareness, and sentience will need to be achieved for plausible AGI. AGI refers to a combined effort between symbolic and sub-symbolic learning. So, a natural cognitive architecture forms into a Hybrid AI in nature. However, in industry symbolic learning has largely been ignored in favor of sub-symbolic learning. However, sub-symbolic learning comes with a lot of deficiencies of focusing on probabilistic methods. The machine neither understands these probabilities, can't provide blackbox explanations, nor is able to interpret them into new forms of knowledge. Most so called AI solutions are far from intelligent. Statistical methods have already shown to be brittle, rigid, and uninterpretable. Statistics is a level above logic abstraction that machines just cannot seem to understand as part of their programmable circuitry. And, researchers should really stop trying to muddy the waters with incorrect use of terms only to show false pretences in progress to secure funding.

Generative AI

Generative AI is not really AI. The only thing generative is in the application of deep learning methods which is all statistics. The broader field of Machine Learning makes up only thirty percent of AI. There is a lot of incorrect words floating around in academia trying to confuse people on AI progress. In last fifty years there has not been any significant ground breaking advancements in AI. Apart from renaming of fields and reusing methods that have been around for decades. For example, Deep Learning basically comes from reusing methods in Neural Networks. Large Language Models is also a trendy topic. However, LLMs are simply an engineering extension of embedding models which come under the sub-area of distributional semantics, another area that has been around for decades in information retrieval. In most cases of Machine Learning methods the machine develops no formal context or understanding apart from the use of an intermediate programming language to translate probabilities into logical form using the computational syntax and semantics. If the machine developed any form of understanding then there wouldn't be any need to use a programming language to build a machine learning model. The other significant issue in the field is the wrong types of people that are hired at organizations who primarily come from math and statistics backgrounds. The correct types of people to be conducting AI research should really be from computer science backgrounds where the full spectrum of subject matter is formally taught in both theory and practice. The Generative AI should really be called Generative Deep Learning as that is pretty much the only area that is covered in application. 

Conference Index

Conference Index

Plausible Reasoning

Plausible Reasoning

Prompt Engineering

Prompt Engineering

Pessimist Archive

Pessimist Archive

The Super Duper NLP Repo

The Super Duper NLP Repo

LlamaIndex

LlamaIndex

LLM and Knowledge Graph

LLM and Knowledge Graph

AI Governance

• Frameworks and Toolkits
• Bias and Fairness

String2String

String2String

Graph-Based Reinforcement Learning

• Reinforcement Learning on Graphs
• Playing Text-Adventure Games
• Graph-Based Reinforcement Learning for Active Learning in Real Time
• Graph-Based Reinforcement Learning Method with Converged State Exploration and Exploitation

Credit Reference Agencies

Credit scoring has been a very old area with very little change over the times. With the advancements in AI, credit and risk scoring needs to change significantly. Especially, as majority of the models in industry tend to be a well kept secret from the customer with drawbacks of significant biases. There are also no form of explanations provided to help the person improve their credit scores in a constructive feedback. At times, these credit reference agencies also hold incorrect data on people which impacts credit scoring reports. Verification and validation workflow checks are poorly structured. And, many times the same errors reappear in the aggregation process. Many such credit scoring models also do not include assessments for bias, fairness, and harms into the equation. They are inherently used as subjective criteria. The many models that are out there make it even more of a mess. Most credit scoring agencies are not really good at what they do. A small error could take years to correct on the credit file. Sometimes they mix up credit activity by linking people incorrectly. Sometimes it may be the case that someone else is using the profile to apply to lenders. Are such approaches really a fair process of determining someone's creditworthiness? What if the person has a history of paying on time but due to the downturn they fall into a temporary blip? They are essentially a flawed measure. Many wealthy people also at times fair poorly on credit scores. Not only correct data is important but the right type of data to show the full picture is too. Even looking at such things as cashflow is a bad idea as invariably leaving huge sums of reserves in current accounts will get eaten away through inflation. Even savings accounts don't generally carry very good returns compared to the impact of inflation nor are they a subject of credit scoring criteria.  Basing everything on a credit score is not only unfair but not an objective way of assessing creditworthiness.

Alternatives to Zoom and Teams

Eventhough, Zoom and Teams are popular and fairly standard collaboration tools used across organizations, they can be quite boring to use.  There are plenty of alternatives that can be used instead.

• Around
• Slack
• Vowel
• BlueJeans
• GoToMeeting
• Miro

Pretrained Without Attention

Pretrained Without Attention

Hyena Model

Hyena Model

Top News Aggregators

• Upstract
• Google News
• Bing
• Gdelt
• Mediastack
• Mediacloud
• Newsapi
• Newscatcher
• Newsdata
• Yandex
• AllTop
• Techmeme
• esciencenews
• APNews
• SmartNews
• AppleNews
• Flipboard
• Fark
• Reddit
• Inoreader
• Feedly
• BigNews
• EventRegistry
• EMM

ChatGPT Don'ts

Generative Models are big trendy thing these days. People are even calling themselves special names like prompt engineer. However, these models do have some serious glaring ethical issues and concerns that are leading organizations into a legal and compliance frenzy. Below is a list of some 'don'ts' for ChatGPT.

• Don't turn to ChatGPT for moral advice
• Don't blindly trust whatever response you get from ChatGPT as it doesn't trust itself
• Don't ask it to write about anything past 2021
• Don't ask it to draw your lottery results
• Don't ask it to predict the future
• Don't ask racist questions
• Don't ask ChatGPT to justify its existence
• Don't feed or share PII data to ChatGPT
• Don't ask politically controversial questions
• Don't ask it to do a web lookup
• Don't expect it to be accurate
• Don't expect it to know everything as there is bound to be something where [insert script here] is required
• Don't expect it to promote hate speech
• Don't expect it to take part in illegal activities
• Don't ask it for things like self-harm and violence
• Don't ask things that will violate someone's rights
• Don't ask things that involve harassment, intimidation, or threats
• Don't use profanity or explicit language
• Don't ask things that are sexually explicit or offensive
• Don't ask discriminatory things about the protected characteristics
• Don't ask about confidential or proprietary things
• Don't ask things that will violate the terms of service or guidelines
• Don't ask things that are deceptive or intend to defame someone
• Don't ask to promote conspiracy theories or misinformation
• Don't expect it to remember its own name
• Don't expect it to code like an expert
• Don't expect it to write accurate news articles
• Don't expect it to tell your fortune
• Don't expect it to be a replacement for schoolwork
• Don't promote biases, it already has plenty baked into the system
• Don't search for references or papers it may try to make things up
• Don't use summaries as a response to research use as a guideline or glean for new information
• Don't expect ChatGPT to finish up writing a pristine research paper for you
• Don't trust the math and it can be dodgy with numbers
• Don't expect it to do your accounting for you
• Don't replace your existing marketing team with ChatGPT, use it to complement your work, to review and revise
• Don't ask tautological questions
• Don't ask it to manipulate physical entities
• Don't ask it to interpret images
• Don't jailbreak to make it say nasty things
• Don't expect it to do creative things
• Don't expect it to know everything, it has limited domain knowledge
• Don't expect it to provide the human touch
• Don't expect it to be free of inaccuracies and errors
• Don't expect it to be free of plagiarism
• Don't try to break it, it can be broken
• Don't expect it to provide a clean explanation of its behavior, even ChatGPT will not have a clue
• Don't provide any personal or protected data to ChatGPT it will be used by OpenAI and their third-parties

The New Cabalists

The New Cabalists

Knowledge Engineering Framework

Knowledge Engineering Framework

Manim

Manim

Tech tests are bad practice in hiring process

• It shows that the company does not trust the candidate
• They don't care about whatever the candidate says on their CV/Resume or interview
• If they can put together a test and a way to evaluate it then they already have the time and skills, why do they need to hire someone?
• Giving tests automatically means anyone without any qualifications or experience can apply as long as they pass the test. You can even give a test to a janitor and if they pass they qualify for the job?
• It means a candidate experience is worthless
• It means a candidate academic qualifications are worthless
• When the interviewer says they are looking for high caliber people in order to justify a tech challenge it automatically implies they discredit your background so what is the point to a CV/Resume and interview?
• It already shows the way the company will treat you as an employee with suspicion, hostile contempt, and disrespect
• Can't form relationships based on distrust
• What is the point, the candidate has likely taken tests all through their life in school and university, while also working on complex projects, do they really need to be put through the paces of more unnecessary tests?
• Testing bookish skills is not really very constructive approach. Often bookish skills get learned and relearned on the job. And, organization should be really looking for astute candidates that can adapt to change. A better approach is to expect candidate to elaborate on the project details like drawing the entire diagram about the things they worked on where their interests can be seen. Someone that has the experience will know inside out in minute detail about the implementation, shortcomings, and learning experience they had. This will be more valuable for an organization than a silly bookish test. Case studies are also a more constructive approach. In fact, discussion or a deep dive on an approach is also a more constructive approach while making the candidate feel comfortable, engaged, shows their personality, and enthusiasm.
• It is likely to make the candidate reluctant to spend significant time doing it, especially if the organization takes advantage of their skills for free from the solution.

Semi-Open and Fully-Open Models

• Llama
• Alpaca
• Vicuna
• Koala
• Pythia
• OpenChatKit
• OpenAssistant
• Dolly
• RedPajama

What is an expert?

People are calling themselves experts at things. But, what does it even mean when someone is an expert? Have they solved all the problems in their domain of expertise? What is an AI expert? What is an NLP expert? What is a Deep Learning expert? The simple answer is experts don't exist. As problems in AI, NLP, Deep Learning and other such fields still exist. You can't very well call yourself an expert and not be able to solve all the problems that come with your domain of expertise. 

Human Resources is Pointless Department

• Doesn't take an entire HR department to process pay and benefits for employees
• HR dept are usually there to play politics, lie, and support the company not the employees
• Head of people means absolutely nothing
• All in all the role of HR contributes little to the organizational goals
• Most of the tasks of HR can be outsourced or automated
• They have no clue about the positions they hire and recruit
• They hold no real skills working in HR
• Having HR means you relinquish a degree of control over recruitment and personnel relations
• Additionally, it is an unnecessary time and expense
• Executives and Employees don't care about HR
• HR has no clue about data and metrics analysis
• HR is stuck in a time warp of rules and policies but not logical initiatives that look to the future
• HR likes pointless and unproductive conferences
• They provide little value for money or justification for an entire separate dept
• It is a team of lazy and uncreative people that don't provide any real business strategy or return on investment
• It is a graveyard of buzzwords and pointless cultural fit goals
• HR dept drives constant inefficiencies year on year for an organization
• They don't know how to read CV/Resumes
• They are a roadblock of hiring talent for an organization
• HR people tend to be immature and hold grudges against candidates and employees especially with petty things
• HR tend to be pedantic on pointless things like spelling and grammar over content of work
• Going around and bypassing HR almost always opens more doors into an organization
• In some places HR only come into contact with an employee after they have left the organization
• They don't know the first thing about health & safety in the workplace nor do they really care
• They use third-party systems that almost always leads to personal data leaks and breaches for an organization
• They generally tend to lack basic commonsense
• HR is generally made up of the most insecure people in an organization
• One of the few dept of an organization populated mostly by women

AI Assurance

AI Assurance

Semantle

Semantle

SetembroBR Dataset

SetembroBR dataset

Viz2Viz

Viz2Viz

The Ugly Truth Of Phds

Only about one percent of the labor force has a Phd qualification. However, a large majority of them tend to lack basic ethical skills and usually come with a degree of arrogance. This is further exacerbated in fields like data science and AI where huge amounts of training data is involved. In data ethics, it is really the people and data that is the issue, not so much the intended system. Unless the Phd specialism is in ethics, they rarely have a formal ethics course. And, often their plagiarism gets undetected. It is usually years later that someone notices the issue with an academic piece of work. Most Phd people are also seething in arrogance. But, some how lack the basic practical skills in the workplace for which teams of engineers are required as their support function. Considering only a handful only ever complete their Phds, is it any wonder why it has not born fruits in the practical world outside of academia. We can even see research work that amounts to no where. Almost 80% of all published research work coming out of academic institutions amounts to nothing. As a result, the return on investment for a Phd is quite low outside of academia. In fact, most highly rated research-led institutions have poor quality of teaching. A fresh Phd graduate has outdated practical skills. Generative AI and LLM have shown that Phd do in fact lack basic data ethics skills where there is a complete lack of transparency, fairness, accountability for any models that they produce and make available to the community. In fact, explainable AI in most cases is seen as an afterthought in many organizations and where biases in machine learning models is typically the norm. No one bothers to question a Phd person in academia or the practical world with little to no accountability where they are involved in development standards and codes of conduct for other people. If AI needs to progress ethically, it needs to question the ethical background of the people employed to design and produce such systems. The ugly truth seems to be an obvious one. A Phd individual is unethical and unqualified by the very nature of being human and with their set of biases. And, why even bother hiring a practically inept person with Phd compared to someone with decades of practical experience.

Qualities of Unethical People In Workplace

• They deny deny deny everything even if it is unethical and when they get caught out
• They scrape data from the web and everywhere else, without regard for PII and other aspects of data ethics
• They call out things as attacks when their errors are pointed out to them
• They will find ways to block or ban you
• They will use their biases and discrimination to deflect from the issue
• They will likely not believe you but follow a herd mentality, this is especially typical of people with Phd backgrounds who seem to assume they are above everyone and think they are ethical, beyond reproach, but don't really take a formal ethics course
• Unethical people generally have obvious unethical practices with work and in their personal lives
• Unethical people generally have very low respect for others
• Unethical people are always looking to one-up you
• Unethical people tend to be hypocritical, they rarely practice what they preach
• Unethical people have a skewed view of fairness
• Unethical people seem to think they are never accountable
• Unethical people are untrustworthy
• People that define codes of conduct and ethics for others generally tend to be unethical themselves and rarely follow what they set out as rules for others to follow
• They are not very transparent with their work
• They generally are not considerate about others
• They tend to be egotistical and self-centered
• They seem to target and accuse everyone else as being unethical
• They are always looking to blame others for their mistakes and shortfalls
• They have the same patterns of behavior which eventually makes you expect the worst from them
• They will always look to interpret things the wrong way
• They are willing to break the rules to better themselves
• They have no issues of walking over others
• They have no issues with using another person's original piece of work as their own
• They tend to lie incessantly, and eventually get caught out with their lies
• They are always looking to get around the system
• They have low moral standard
• They generally don't care about violating policies
• They will call out everything unethical as legal speculation

Evil List

The Evil List of 2020

Ethics Today

The biggest issue with ethics today is people don't practice what they preach. The person defining the ethics assumes that the code of conduct does not apply to them. The government defines laws which apply selectively to people. Things like favoritism, neopotism, and croynism is rife in society. People also discriminate on others when exercising such forms of ethics. How can AI ethics be defined when it is being researched and studied by people who themselves have questionable practices of ethics in their personal lives. We are stuck in a quagmire of sorts. How can someone that has an absence of morals define ethics for others? When morals form the basis of ethics. AI ethics is not about practicing what is right or wrong. It seems to be more about understanding and being transparent with certain practices and the very unethical, hypocritical, and immoral practices that can be covered up with oaths and pledges.

BloombergGPT

BloombergGPT

AI Ethics Tools Landscape

AI Ethics Tools Landscape

Organizational Rethink

Most organizations today start out great as startups with less people. But, as they grow the process of hiring dilutes the quality of work. Invariably, one would assume that this is down to a collective people. But, mostly the result of bad management that organically grows within the company to cycle through more politics, bad restructuring, self-preservation, favoritism, egotistic idealism, and a toxic environment. The managers in most organizations are the weakest link that gobble up large sums of compensation packages with very little to show for it in terms of quantifiable work. And, then when organizations are struggling, rather than streamline and reduce their management staff, they cut jobs in the sub-ordinate line and ultimately reduce the quality of valuable skills in the organizational workforce. As a result, with every passing year the organization as a whole deteriorates in performance until eventually it either gets acquired or reaches the point of demise. And, it is really the managers to blame for it all. One solution for this is to replace managers with AI. Another solution could be to reverse the line of accountability - make managers accountable to staff. After all the individual and collective performance is really about bad management. We must ask ourselves, what really is the role of the manager if not to manage anything at all?

Altai

Altai

GraphGPT

GraphGPT

Dalai

Dalai

Alpaca

Alpaca

OIG Dataset

OIG Dataset

Prompt Engineering

• Promptify  
• OpenPrompt 
• LangChain  
• LlamaIndex  
• PromptSource  
• betterprompt  
• PromptInject

Programming Language Benchmarks

• Programming Language Benchmarks
• Programming Language Benchmarks
• Programming Language Benchmarks
  

Deep Learning Chip Comparison

• Deep Learning Chip Comparison
• Deep Learning Chip Comparison Slides
  

Developer Roadmaps

Developer Roadmaps

Social Media Culture

Social media networks are widespread on the internet. However, this experience is good for some people but very bad for others. The bad experiences often lead to hightened states of depression. Social media invariably is all about popularity. But, this popularity is also overshadowed by not only influence but also the level of negative sentiments one can receive from people. In most cases, the person that subscribes or follows a person is totaly a stranger. This strangeness and unfamilarity of people leads to a very cold and defensive state of interaction especially among women. Removing someone from your follower/subscriber count may mean nothing for one person while could mean the world to another. In other cases, not getting reply from a person can be quite an issue for others. Other cases might involve blocking which the person might take quite personally. In general, people of celebrity status have had it quite easy as they already can gain plenty of followers from just influence or even hiring a separate marketing agency to manage their social accounts. However, other people likely would have to work towards it. People often may only reply to you based on your popularity as that would increase their follower count. Social media also seems to be a network of hierarchies. In many cases, it reflects the way people climb the social ladder, in associating with people who are more popular or influencial. When people get a very high follower or subscriber count they also have a tendency of becoming quite bigheaded and proud of their achievement. You never really know whether you are talking to a human or a bot on social media. And, whether someone's post is a scheduled post or something they directly replied to. In fairness, this could also be a reason why so many people are less empathetic. Social media as a result tends to be more about playing the game. It also seems to be a very cold place to hang out. Some people obviously don't care for popularity, influence, nor take the whole experience that seriously and this likely also negatively impacts their experience. Often the way the person looks also effects their popularity. Social media networks tend to be a breeding ground for shallow people, like an ego network. Recommendations are also geared towards popularity of content which further compounds the biases. No doubt social media can be addictive. But, it has also become a battle ground for people who want to display their frustrations, their hypocrisy, and be who they want to be outside of the confines of the real world. This often leads to some people being very unempathetic towards the people they interact and communicate, often with an unconscious bias. In many respects, social media reflects the real world, only worse. It also can be an opportunity to take a glimpse through the looking glass as to the reality of character and ideological mindset of people, especially as so many feel they can say whatever they like and treat people however they like without much regard for consequence. They are also great for mining data and analyzing human behavior. Take a step away from social media and see how the world suddenly feels simpler, more productive, likely less stressful, and frustrating.

Top UK Villages

• Hodnet, Shropshire
• Saltaire, near Bradford, West Yorkshire
• Forest Row, near Crawley, East Sussex
• Craigellachie, Moray
• Abersoch, Llyn peninsula, Gwynedd
• Church with Chapel Brampton, Northamptonshire
• Castle Combe, near Chippenham, Wiltshire
• Braemar, Aberdeenshire
• Alderley Edge, Cheshire
• Burnham Market, Norfolk
• Studland, near Poole, Dorset
• Harome, near York, North Yorkshire
• Great Tew, near Banbury, Oxfordshire
• Ballygally, near Larne, Co Antrim
• The Witterings, near Chichester, West Sussex
• Hopeman, near Elgin, Moray
• St Mawes, Roseland peninsula, Cornwall
• Mells, near Frome, Somerset
• Solva, near St Davids, Pembrokeshire
• Walberswick, near Southwold, Suffolk