Ontologists, Data Scientists, and Researchers fulfill a very narrow role function in organizations. Invariably, they are unnecessary and their functions can be absorbed into capable engineer roles especially ones that have computer science and engineering backgrounds. An ontologist builds ontologies who are incapable of pipelining or building a knowledge graph. A data scientist builds models who are incapable of doing feature engineering, cleaning data, scaling out, or pipelining their own work. A researcher conducts research who are incapable of scaling out their work or following basic software engineering practices. In a way they hack their way through an imperfectly produced artefact and a published paper that usually, in practical terms, does not convert to a quantifiable and significant value for organizations. Not only this, but they also need other engineers to help them with 80% of their work. Ever seen an ontologist getting involved in W3C specifications meetings where it is mostly consumed by engineers and architects with practical experiences of understanding business domains? Many organizations don't even have a need for an ontologist or a taxonomist as the role is being played by an engineer or an architect. Having to build an entire team around such useless individuals is an unnecessary cost especially as they have little to no skills in transfering theory into practice. And, in most cases, their abstract skills are transferrable across other disciplines who may in fact be in a more qualified position to conduct such work activities. This issue is further compounded today with fake job ads where machine learning engineers, nlp engineers, computer vision engineers, deep learning engineers are required while not needing to do any machine learning, nlp, computer vision, or deep learning work. In fact, these engineer roles purely turn into devops functions to support an inadequate and incompetent data scientist who should really be responsible for the entire end-to-end data science method. In many cases this perpetuates recruitment fraud as the job titles don't match the job descriptions in the slightest and in trying to make a fool out of candidates. Such bad practices across industry sectors are only going to get worse through hiring of Phd people that only account for 1% of the population, that have limited practical experiences, while dictating to clueless management on how things should be done, in an effort to save face with their inadequacies of fulfilling their full role responsibilities.
23 April 2022
19 April 2022
18 April 2022
Online Fact-Checking
The coverage of ukraine-russia conflict has shown that online fact-checking is not only farcical, in variably filled with biases, and often manipulated by regional governments. To evaluate for credibility for trust one has to start with an objective view in analyzing a balanced set of sources of input in order to garner a full picture of claims and biases before processing for a set of filters in disinformation and misinformation. This implies looking at both sides of the argument. However, as pro-russian media outlets are blocked and only pro-ukrainian western mainstream media is allowed coverage, a one-sided narrative is provided to the masses. As a result, the credibility process becomes flawed which undermines the process of fact-checking. The web has become a pre-filtered echo chamber where blocking freedoms of speech are ruthlessly defended through geographic jurisdictions. The western notion of freedom of speech emerges as a facade. In a court of law, there tend to be two-sides, a defense and a prosecution. On the web and social media, the defense is blocked and only a biased prosecution is made available. If a government jurisdiction can intervene to suppress freedoms of speech then it leaves little in process of credibility evaluation, especially when the concluding evidential claims is found not to fit or support the narrative. What if the mainstream coverage is an inaccurate projection of events happening on the ground, that would surely result in a cascade of fake news being treated as credible and supported by regional government narratives. In order to accurately evaluate and deliver on credibility one has to go beyond the fold.
14 April 2022
9 April 2022
Dialog System Toolkits
Codified Toolkits (No coding required):
- Botmock
- Botpress
- Botsify
- Botsociety
- Chatbase
- Chatfuel
- Chatlayer
- Dashbot
- Engati
- Flow.ai
- FlowXO
- Gutshup
- Hubspot
- Manychat
- MobileMonkey
- Morph.ai
- Octane AI
- Sequel
- Voiceflow
Scripting Toolkits:
- BotKit
- Chatscript
- PandoraBots
Commercial Toolkits:
- Alexa Skills
- Lex
- DialogFlow
- Facebook Messenger
- Haptik
- Watson Assistant
- Bot Framework
- Digital Assistant
- Rasa
- Conversational AI
- Teneo
- Wit.AI
Research Toolkits:
- ConvLab
- DeepPavlov
- Olympus
- OpenDial
- ParlAI
- Plato
- Pydial
- Virtual Human Toolkit