How to build a Big Data Analytics system based on artificial intelligence more perfect than ChatGPT that learns but only real information and data?
How to build a Big Data Analytics system, a Big Data Analytics system, analysing information taken from the Internet, an analytics system based on artificial intelligence conducting real-time analytics, integrated with an Internet search engine, but an artificial intelligence system more perfect than ChatGPT, which will, through discussion with Internet users, improve data verification and will learn but only real information and data?
Well, ChatGPT is not perfect in terms of self-learning new content and perfecting the answers it gives, because it happens to give confirmation answers when there is information or data that is not factually correct in the question formulated by the Internet user. In this way, ChatGPT can learn new content in the process of learning new but also false information, fictitious data, in the framework of the 'discussions' held. Currently, various technology companies are planning to create, develop and implement computerised analytical systems based on artificial intelligence technology similar to ChatGPT, which will find application in various fields of big data analytics, will find application in various fields of business and research work, in various business entities and institutions operating in different sectors and industries of the economy. One of the directions of development of this kind of artificial intelligence technology and applications of this technology are plans to build a system of analysis of large data sets, a system of Big Data Analytics, analysis of information taken from the Internet, an analytical system based on artificial intelligence conducting analytics in real time, integrated with an Internet search engine, but an artificial intelligence system more perfect than ChatGPT, which will, through discussion with Internet users, improve data verification and will learn but only real information and data. Some of the technology companies are already working on this, i.e. on creating this kind of technological solutions and applications of artificial intelligence technology similar to ChatGPT. But presumably many technology start-ups that plan to create, develop and implement business specific technological innovations based on a specific generation of artificial intelligence technology similar to ChatGPPT are also considering undertaking research in this area and perhaps developing a start-up based on a business concept of which technological innovation 4.0, including the aforementioned artificial intelligence technologies, is a key determinant.
In view of the above, I address the following question to the esteemed community of scientists and researchers:
How to build a Big Data Analytics system, a system of Big Data Analytics, analysis of information taken from the Internet, an analytical system based on Artificial Intelligence conducting real-time analytics, integrated with an Internet search engine, but an Artificial Intelligence system more perfect than ChatGPT, which will, through discussion with Internet users, improve data verification and will learn but only real information and data?
What do you think about this topic?
What is your opinion on this subject?
Please respond,
I invite you all to discuss,
Thank you very much,
Best wishes,
Dariusz Prokopowicz
First you have to create a system that will be resistant to various data structures, and it turns out it should not be a neural network.
Dear Dariusz Prokopowicz ,
In the last years, more and more managers in organizations are reliant on a wide range of information systems that provide them with analytics methods and features to support their decisions and planning activities. With the trend of the growing automation of business processes and subsequently growing a huge amount of data stored in databases, big data analytics methods can now take full advantage of this trend to support decision makers in their decision-making process.
Article DEVELOPING CHATBOT SYSTEM TO SUPPORT DECISION MAKING BASED O...
Regards,
Shafagat
Dariusz Prokopowicz ,
This is a very complex question but I will try to synthesize my main points into what I consider is the main problem with LLMs and my perceived solution.
One of the underlying problems with LLMs is the problem of hallucinations and the wrong answers it generates. This has its roots on two subproblems. The first is the data and its training, the second is in the nature of the algorithms and the assumption of graceful degradation. I think that the first one is easy to solve by not throwing junk data and expecting that 'statistical miracles' occur and bubble up truth from noise. That is a nice mathematical hallucination on our part (no amount of mathematical Platonism can compete with the messy "mundane" day to day ). There is no replacement for hard work to sort out good data from bad one.
The second problem is the one that is more difficult to solve. It lies on several assumptions that are ingrained in neural networks. Neural networks promised graceful degradation, but in reality we need neural networks to abstain from graceful degradation in critical situations. Hallucination is based on this philosophical flaw of neural networks. The graceful degradation relies on distributed representations and the assumption that even thought the whole representation is not present, if there is enough of a representation it will output the complete representation. This is an extremely strong assumption to embrace as a universal case for all data. This is by necessity an existential case and not a universal one. A possible solution to this is to use an ensemble of algorithms that contain neural and non neural algorithms and the consensus wins.
In my view, both curation of primary data for foundational models and the consensus of algorithms is necessary (but not sufficient) to achieve a better system. I would also tackle how to realize these two solutions as a separate thread for each one.
Regards
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation