The current industry buzz surrounding Artificial Intelligence has been a long time in the making. The first wave of Artificial intelligence during the 60’s commenced with the implementation of systems like CLIPS and other expert advisors. It was a let down in terms of the deliverables produced by the academics rousted into industry.
The failure of the first wave of AI can be explained by a number of factors;
- Lack of understanding about the problem space. Computer science is still a relatively young field, and it is easy to forget that the very first personal computer was released in 1964.
- Computation issues. E.g. Neural nets where first implemented by Warren McCulloch and Walter Pitts in 1943! The mathematical models and algorithms were conceived more than 70 years ago. However, due to the highly resource-intensive computations required to perform the training of the nets, this was disregarded at the time.
- Implementation issues – the systems were often difficult to create and required highly specialised hardware, software and experts to both implement and maintain.
As further generations of AI researchers worked in the field of artificial intelligence, the theoretical progress was fast, from Rosenblatt’s Perceptron through to Hebbian Learning Nets. However, computation was still lacking and there was no real commercial applications to the technology
Moore’s Law (computation power will roughly double every 4 years) has put paid to the 2nd issue. 70 years ago, these researchers still had to wire up their own hardware by hand. Now, the computation power contained within a pocket calculator is greater than that of the shuttle which landed Neil Armstrong on the Moon. Computation is no longer an issue.
As a data driven organisation, Agent3 is fully committed to leveraging cutting edge technologies and techniques. John Clark said it best;
“The intelligence of AI is often interpreted as mirroring human capabilities, but the scale of data potentially … places analysis well beyond human capabilities.”
We firmly believe that AI is a rising wave, and as the ability of computers to perform ever more complex tasks, investment and resources will be piled into this area. This means that without a constant focus on innovation and the applications of this revolutionary technology, companies risk being left behind by those who have recognised the strategic importance of heavy investment in AI. In our quest to find solutions to the third issue above, we have been exploring providers such as AWS.
Cloud providers such as AWS provide the facility to deploy advanced techniques in an effective manner. This reduces the need for specialised hardware/software/experts and with the use of services such as Sagemaker organisations today are able to deploy the same code used throughout prototyping/development to production.
As can be seen, this allows an organisation to dramatically reduce the amount of time taken to move a new model to production.