Description of why this would be an interesting post
The field of AI + IoT (AIoT) in recent years has without a doubt been growing in popularity. I’ve done some background research on a topic that we call AIoT Blockchain, and the research team in my lab has recently completed our newest work on this topic: MLAV (Multi-Layer Aggregrate Verification).
This paper gives a non-technical and theoretical description and forecast of the interaction between AI, blockchain, and IoT. Non-technical studies such as these may have major contribution and play crucial roles in allowing for researchers to access the true potential of more technical ones.
The larger the quantity of data produced by IoT devices, the better the AI’s performance will be. IoT devices naturally have limited data storage capabilities, and this is precisely where blockchain’s scalability plays a role of key importance.
Thanks for the reply Lucas,
In my experience, most works and commentaries that mention the overlap between AI and blockchain tend to go in one of three directions: smart contracts (rule-based system regulations, neural graphs, neural networks, etc.), certainly trading bots as you mentioned, and also blockchain as a storage mechanism for AI models. I’d love to do more research in each of these directions, and in terms of further research after this paper, AI in trading bots would definitely be a good place to start.
Sounds like a great research path – excited to read your summary on Convergence of Blockchain, IoT, and AI, it’s a good starting point for coverage in this area.
I saw you posted in the other thread that @rncd created about autonomic computing models, so it might be beneficial to collaborate on research in that both parties have an interest in automated modeling. Not to say that is the only way to approach doing research on this subject, but I have found that checking with other researchers who have a similar interest tends to accelerate the work of all collaborating parties.
This is a good idea to examine a combined study of all of these technologies. We are in a great need for advanced technologies to securely handle and efficiently process big data. Looking at some current trends:
GPT-3 enabled on Microsoft Power Fx has the capability to generate a program through the power of NLP.
IBM Hyperledger Fabric platform may boost the practicality of blockchains.
Yes, if you look at Google IO this year, they announce Multitask Unified Model (MUM) for search. GPT-3 or BERT might be the one of the engines beyond that service, but ultimately, the trend seems to make it more intellegent…, doesn’t know if there is a better word for that.
Interesting read. I found the use case of turning public assets into blockchain-based AIoT devices particularly appealing. Setting up “smart” tire inflators, tokenizing their ownership on a blockchain, and selling the tokens to local investors, for example, would make for a fascinating social experiment.
@rncd I remembered you had an excellent thread on here about AI and I wondered if you might be interested in weighing in on this idea for another summary
I just read through this article Lucas sent, and I thought I might as well put my own notes here, first to give you a head start on your summary, and second to enhance my background knowledge of interdisciplinary blockchain AI before I finish the AIoT Blockchain summary.
This Stanford article highlights the fact that the processes in machine learning are frequently cost-inefficient and difficult to access for many. Datasets are frequently privately-owned and expensive to create. Therefore, it’s all about using blockchain to democratize AI.
Blockchain offers trust and security. Through smart contracts, it also offers reliability in execution of code and transparency. Smart contracts are specifically what is used for contribution of data. Hosting an ML model requires a one-time fee for uploading and a small fee for anyone uploading data. The authors propose three methods for incentivizing people to upload “good” data: gamified (rewarding based on validation and acknowledgement from other contributors), prediction market-based (reward based on how the data submitted affects the model’s accuracy), and ongoing self-assessment (reward based on mutual validation and mutual payment for good data).
Right now, this framework is most useful for smaller models, and research in scaling still needs to be done. Overall, however, a promising area of research!
Hey @Gearlad. Great post. I’m actually actively involved in this research area myself (Blockchain + IoT) and there are definitely novel engineering challenges involving consensus, governance, etc. One of the fundamental limitations is how you have a performant + scalable enough consensus algo that doesn’t skimp too much on fault tolerance.
Most Blockchain (really DAG-based DLT) projects I know that can be adapted to IoT use some flavor of PBFT (Practical Byzantine Fault tolerance), but it has subpar fault tolerance threshold at ~33% compared to PoW (Proof-Of-Work) 51%. I have linked a paper that may be useful to you regarding this
According to the article, I realized that Blockchain can compensate for the shortcomings of the current artificial intelligent field. With the development of this field, the model dramatically increases in parameter, there for the size taken to store or run the model. Take the well-known GPT3 for an example, the model would require at least 350 GB of VRAM just to load the model and run inference at a decent speed. Therefore, individuals can’t run the model, but with the combination of the blockchain, it might be possible for everyone to take advantage of artificial intelligence.
Additionally, just like the article mentioned, most of the dataset is privately owned by big companies. Most people couldn’t train their own model in the topic they are interested in. For example, most medical datasets are owned by either institution or hospital, which isn’t public. Only big companies can access or generate the dataset they want to train their own model and then charge users using the service they provide. Which create the dilemma that small company and individual can never harness the full power of artificial intelligence, and the big companies can take advantage of artificial intelligence to gain themselves. However, with the incentive mechanisms, the participant can contribute their data to the blockchain, by aggregating the data everyone provides. There will be a fair amount of data enables everyone to train model related to the topic they like, which will be wonderful.
Yes, there have been many papers published that explore the use of artificial intelligence (AI) in the context of bots, which are computer programs that are designed to automate tasks or interact with users online.
Here are a few examples of papers that discuss AI and bots:
“Bots and AI: The Future of Online Advertising” by J.C. Spink and B.J. Fogg discusses how AI is being used to improve the effectiveness of online advertising through the use of bots. The paper discusses the benefits and challenges of using bots in online advertising and provides recommendations for best practices.
“The Role of Artificial Intelligence in Chatbots” by J.H. Lee and J.H. Kwon examines the use of AI in chatbots, which are bots that are designed to engage in conversation with users. The paper discusses the various types of chatbots that use AI, the benefits and limitations of these chatbots, and the ethical considerations that must be taken into account when using AI in chatbots.
“AI-Powered Bots and the Future of Customer Service” by B.J. Fogg and J.C. Spink discusses how AI is being used to improve customer service through the use of bots. The paper discusses the benefits of using AI-powered bots in customer service, as well as the challenges and considerations that must be taken into account when implementing such systems.
Overall, AI has the potential to significantly improve the capabilities of bots, enabling them to perform tasks more effectively and engage with users in more natural and intuitive ways. However, it is important to carefully consider the ethical implications of using AI in bots and to ensure that these systems are implemented in a responsible and transparent manner.