Blockchain/ Distributed Ledger

So, looking forwards, Generative AI is getting really good at creating digital work. We're exploring what this means and some possible strategies for dealing with this.

Long term this leads to some sticky questions about the creation and consumption of AI generated works (and that's ignoring legal issues like copyright, we'll skip that for now). Long story short: bad content can waste your time and even lead you down paths that can result in failure, as an extreme example there is recently a trend of AI generated mushroom foraging guides, to put it bluntly there's a LOT of poisonous and dangerous mushrooms, pick and eat the wrong ones and you could end up sick or dead (https://www.google.com/search?q=ai+generated+mushroom+books). 

So with this in mind I've started exploring the space, we had a really productive meeting today that can be summarized roughly as:

• Building trust that the content is good and correct, that it has quality and value and is with consuming.
• Transparency, what parts of the digital work (e.g. a paper) were generated by people, by AI? By both (collaboratively?)
• What processes, models, inputs and outputs were used with respect to the AI generated content?
• Why was AI used, e.g. for the Generative AI Policy paper for some sections we used AI to convert the ideas and arguments put forth by people and the raw feedstock (e.g. the meeting minutes, transcripts and so on) into written text, one advantage of this is as we refine our ideas and arguments we can quickly rewrite the paper (which I will already need to do because of massive movements and changes in the last 2 weeks), because we used AI we can rewrite the paper much more quickly and accurately.
• Providing information about the possible bias and discrimination of both the AI and the humans working on the paper (e.g. meetings were held in english, in the north american timezone, so that creates some inherent bias)
• Using blockchain to collate and provide the evidence, e.g. Working Group calls, emails, Circle discussion, drafts, and so on all have their own logging and proof (e.g. Google docs history, email that is DKIM signed, and so on) but this needs to be packaged up. This is not only useful for proving where the content came from, but for further developing it and updating it.

So, looking forwards, Generative AI is getting really good at creating digital work. We're exploring what this means and some possible strategies for dealing with this.

Long term this leads to some sticky questions about the creation and consumption of AI generated works (and that's ignoring legal issues like copyright, we'll skip that for now). Long story short: bad content can waste your time and even lead you down paths that can result in failure, as an extreme example there is recently a trend of AI generated mushroom foraging guides, to put it bluntly there's a LOT of poisonous and dangerous mushrooms, pick and eat the wrong ones and you could end up sick or dead (https://www.google.com/search?q=ai+generated+mushroom+books). 

So with this in mind I've started exploring the space, we had a really productive meeting today that can be summarized roughly as:

• Building trust that the content is good and correct, that it has quality and value and is with consuming.
• Transparency, what parts of the digital work (e.g. a paper) were generated by people, by AI? By both (collaboratively?)
• What processes, models, inputs and outputs were used with respect to the AI generated content?
• Why was AI used, e.g. for the Generative AI Policy paper for some sections we used AI to convert the ideas and arguments put forth by people and the raw feedstock (e.g. the meeting minutes, transcripts and so on) into written text, one advantage of this is as we refine our ideas and arguments we can quickly rewrite the paper (which I will already need to do because of massive movements and changes in the last 2 weeks), because we used AI we can rewrite the paper much more quickly and accurately.
• Providing information about the possible bias and discrimination of both the AI and the humans working on the paper (e.g. meetings were held in english, in the north american timezone, so that creates some inherent bias)
• Using blockchain to collate and provide the evidence, e.g. Working Group calls, emails, Circle discussion, drafts, and so on all have their own logging and proof (e.g. Google docs history, email that is DKIM signed, and so on) but this needs to be packaged up. This is not only useful for proving where the content came from, but for further developing it and updating it.

------------------------------
Kurt Seifried
Chief Blockchain Officer and Director of Special Projects
Cloud Security Alliance
[email protected]
------------------------------

Recently I had an interesting experience with ChatGPT.   I was tasked to develop a one pager product data sheet for a startup I advise.   I asked the CEO to send me his top value proposition messages.   Instead he sent me a ChatGPT generated one pager, that probably scanned the website and other literature or competitive literature on the net.   It was 90% correct and really cut my work by at least 50%.  However a few sentences were either incorrect (questionable or non sequitur) or irrelevant for our market or target customers.  ChatGPT probably had tapped into information that appeared to relate to what we were doing, but was not.  Bottom line, ChatGPT is  risky in the hands of individuals who either lack common sense, critical thinking, or simply knowledge of the their field.   But for individuals that have the knowledge, it is an incredible time saving.   It adresses the "blank page" syndrome, and changes your task from writing to reviewing .  Much simpler task if you have the knowledge and  are not a writer by trade.

So, looking forwards, Generative AI is getting really good at creating digital work. We're exploring what this means and some possible strategies for dealing with this.

Long term this leads to some sticky questions about the creation and consumption of AI generated works (and that's ignoring legal issues like copyright, we'll skip that for now). Long story short: bad content can waste your time and even lead you down paths that can result in failure, as an extreme example there is recently a trend of AI generated mushroom foraging guides, to put it bluntly there's a LOT of poisonous and dangerous mushrooms, pick and eat the wrong ones and you could end up sick or dead (https://www.google.com/search?q=ai+generated+mushroom+books). 

So with this in mind I've started exploring the space, we had a really productive meeting today that can be summarized roughly as:

• Building trust that the content is good and correct, that it has quality and value and is with consuming.
• Transparency, what parts of the digital work (e.g. a paper) were generated by people, by AI? By both (collaboratively?)
• What processes, models, inputs and outputs were used with respect to the AI generated content?
• Why was AI used, e.g. for the Generative AI Policy paper for some sections we used AI to convert the ideas and arguments put forth by people and the raw feedstock (e.g. the meeting minutes, transcripts and so on) into written text, one advantage of this is as we refine our ideas and arguments we can quickly rewrite the paper (which I will already need to do because of massive movements and changes in the last 2 weeks), because we used AI we can rewrite the paper much more quickly and accurately.
• Providing information about the possible bias and discrimination of both the AI and the humans working on the paper (e.g. meetings were held in english, in the north american timezone, so that creates some inherent bias)
• Using blockchain to collate and provide the evidence, e.g. Working Group calls, emails, Circle discussion, drafts, and so on all have their own logging and proof (e.g. Google docs history, email that is DKIM signed, and so on) but this needs to be packaged up. This is not only useful for proving where the content came from, but for further developing it and updating it.

------------------------------
Kurt Seifried
Chief Blockchain Officer and Director of Special Projects
Cloud Security Alliance
[email protected]
------------------------------