We recently challenged our SkyBOXX customer community to read my three previous Deep Learning blog posts and respond with up to five questions regarding Deep Learning/Artificial Intelligence.
The following are answers to the most frequently asked questions.
Q1: Can the Deep learning data or information be copied and simply applied to other Machines?
Yes, it can. With Deep Learning, the training phase trains the system based upon data fed to the neural network. Once training is complete, the next stage is the Inference stage where data is fed to the AI which subsequently uses that training to perform its assigned tasks.
Q2: If the deep learned information is simply copied and pasted to other machines will they interact with their surrounding as perfect clones or will there be alternations due to their environments if different?
They should perform identically (like at a factory with multiple assembly lines). Alterations in the trained data set should only occur if the system enters training mode.
Q3: Could blockchain technology be utilized to verify or influence deep learning? Furthermore, could that technology be used to expand the available processing power to AI or a network of AIs?
Blockchain and deep / machine learning both require significant amounts of energy for what they do. Blockchain requires the solving of complex mathematical problems for both the encoding process and the decoding process.
Blockchain consumes a large amount of processing power in order to build the blocks in the chain. AI systems consume a large amount of processing power also. However, AI could release some of that power if the data feed to the AI slowed down.
Q4: Will AI be able to take my job as a machinist?
This is a mixed answer. AI will be able to take away some of the job duties, which should give more time to the machinist to focus on other more pertinent tasks. With short term projects, you are likely to find that it is much faster doing it the manual way as opposed to not only having to retool the lines for the new product, but also train up a new neural network.
Not to mention, certain areas in the machinist profession require judgment calls that could be difficult for an AI to handle. Maintenance machinist David Dinwiddie (full disclosure: we are related) says, “That’s what I have been doing for 37 years—making judgment calls constantly. If you get a damaged shaft, you have to decide if you are going to cut it off and replace the whole end or turn it down and put a sleeve over it.”
Q5: With so little trust given between humans, how can trust be given to machines with advanced AI? Even with advanced AI, the computer still cannot think for itself. It is restricted to following the results of the training period as guidelines for sorting out the data it is receiving in the inference stage. It is expected that when combining blockchain technology with an AI system, the AI could achieve levels of trust by utilizing secure blocks on the chain.
Thank you to all who took the time to read the articles and submit their questions.