I do not think that any approach to AI can ignore the massive data provided by the internet, part of which is nothing more than the digitalization of pre-internet or non-internet material. There is of course the problem of the enormously varying quality and reliability of this material, the presence of redundancy and its sheer vastness, which could lead one to wonder whether processing such raw data via rudimentary algorithms is really worth the energetic and environmental costs or the use of the expensive infrastructure involved.
I believe that the correct approach to AI must be based on formal logic and the logical-algebraic frameworks of theoretical computer science, as well as other kinds of mathematics beyond the the ones commonly employed in machine learning.
The Semantic Web project seemed a good approach along these lines. It involves a logical and formal semantic analysis of natural language. It calls for a far more sophisticated way of producing internet content and (re)presenting human knowledge on the internet. No Data without Metadata. We need a machine-human logical-semantic interlingua so that internet data can become machine readable in a logical and semantic sense (rather than mere statistical data chunked by a machine learning algorithm).
We should be able to effect complex structured queries to intelligent evolving self-correcting interlinked data bases according to varying degrees of precision which will be able to output the source and a measure of reliability of the data presented.
Machine learning will come into play for example at the level of automatic theorem proving, of the massively difficult task of the processing of logical queries.
Our ethical principles can be given formal logical formulation than can be understood by machines.
It seems that this approach (even if demanding more time and work and being filled with challenges) is far more desirable than internet-based Large Language Models. This kind of 'intelligent' AI seems to be in the long run a better ethical , environmental and human choice.
Your perspective emphasizes the importance of a formal, logical approach to AI, leveraging semantic analysis and structured data. It raises valid concerns about the reliability and environmental costs associated with processing vast amounts of internet data using rudimentary algorithms. The Semantic Web project's emphasis on logical and formal semantic analysis aligns with your vision for a machine-readable, structured representation of knowledge. Balancing ethical considerations and environmental impact in AI development is crucial, and your preference for a logical-semantic interlingua for complex queries reflects a commitment to a more refined and sustainable approach.
Dear Clarence Lewis Protin ,
The Internet of Things (IoT) has continued to be a key development in the tech industry throughout 2017, with many businesses looking to utilise IoT within their key processes and daily tasks. Many have also integrated the use of artificial intelligence and big data as a result – but what does it all mean for the future?
What are IoT, big data and AI?
Before we delve into how this tech trinity can benefit your business in the present and for the foreseeable future it is important to gain a basic understanding of what these three elements are.
IoT
The Internet of Things is an IT network of devices, ranging from the smartphone on your desk through to machinery and buildings, all the devices are connected – making up The Internet of Things. Each device is fitted with sensors, known as ‘dumb’ sensors, which collect huge amounts of data each minute of the day.
Big Data
Big data is simply the information that these connected devices gather over time. There is much a vast amount of information that many businesses are failing to harness the true potential of big data and how it can benefit their business.
https://www.seriun.co.uk/iot-big-data-ai/
Regards,
Shafagat
I agree with you, Clarence Lewis Protin, on your preference for more carefully crafted AI over internet-scale models to ensure ethical, sustainable progress.
I'm yet to experience an Intelligent AI that's able to generate text or imagery based propositions with the relevant sources and the actual links to the sources that these texts or images were generated from. As far as we humans are concerned we build on existing contents and inventions, and make relevant references to prior state of the art in the field. However, these AIs available has such serious limitations on acknowledgement of their extracted sources where their contents are generated.
Moreso, it is of importance that Intelligent AI irrespective of how algorithmically sound it may be, it should be trained to be self-correcting and self-checking of facts as it generates its contents. For the reason, that plagiarism and ethical issues are of the essence that if these intelligent AIs should be developed in such a manner to incorporate these concerns to produce ethically sound arguments and contents that can be verifiable by interested parties.
These limitations make the current state of AIs not at par for establishing a standard in comparison to the academic community and as well to the creative communities. If these bias are addressed through well crafted AI, then AI will become the most powerful productivity tool in the hands of we humans.
- Badges
- Science method