Re: AI KR Strategist core plan

Paola

Thanks for your comments .. to ensure your points are addressed in a manner
you think appropriate  - please propose possible edits which will be
reviewed at our next call.

The 'plan' is not attempting to be a white paper so references are not
being provided. It would be helpful if you can identify where they might be
used so we can improve the explanatory description

cheers
carl
It was a pleasure to clarify


On Thu, Jul 23, 2020 at 11:18 PM Paola Di Maio <paoladimaio10@gmail.com>
wrote:

> Carl
>
> Thanks again for advancing a plan for an AI KR strategy
>
> some of is clear and make sense to me, but some of I find it hard to
> digest and would like to know where did you get some of these ideas from?
> They dont seem to belong to AI KR as I know it, perhaps I need some
> clarifications
>
> ------------------------------------------------------------------------------------------------------
> In relation to my suggested goals
>   please highlight  in the document that you have sent the measurable
> properties for the goals which are already stated (I can see the algorithm
> evaluation ones but not others)
> I ll get back to you for some metrics for my proposed goals
>
> -------------------------------------------------------------------------------------------------------------
>
>
> Having taken a look at the plan so far, it looks that some of the goals
> are not clearly state, for example - risk -  should the goal be risk
> aversion?
>
> Overall, I think some of these elements could be optimised/better
> organised but need to think
>
> I wonder if it would be beneficial to distinguish between Natural Language
> and ML KR somewhere in this plan, looks like the two are bit mixed . They
> can mix like in neurosymbolic integration  but thre should be a rationale
> and a clear method for doing so
> Finally, I really think there should be references and perhaps some
> justification
>
> For example in algorithmic evaluation performace there are
> some techniques listed which I cannot understand where they come from, and
> how they are applicable to what we are doing here
>
> If you take for example the algorithmic evaluation overview by NIST
> none of those techniques are mentioned
>
> https://www.nist.gov/system/files/documents/calibrations/5366.pdf
>
> May I ask whats that stuff and is in there (I dont think it belongs there
> in any meaningful way)
>
> Finally, where is the list below come from? (please reference the source)
>
>  an Intelligent Reasoning (fragmentary) Theory, such as,
> o deduction,
> o induction,
> o abduction,
> o by analogy,
> o probabilistic,
> o case-based
>
>
>
>
> On Thu, Jul 23, 2020 at 9:46 PM carl mattocks <carlmattocks@gmail.com>
> wrote:
>
>> Paola
>>
>> agreed - it would be helpful if your explanations included measurable
>> properties
>>
>> thanks
>> Carl
>> It was a pleasure to clarify
>>
>>
>> On Thu, Jul 23, 2020 at 12:01 AM Paola Di Maio <paola.dimaio@gmail.com>
>> wrote:
>>
>>> Carl
>>> I have seen bias on the sheet
>>> I d like to add
>>>
>>> logical and functional system integrity
>>> semantic precision and consistency
>>> preservation of truth values
>>>
>>> Let me know if any of these need explaining
>>>
>>> could you please add those in as well at the next opportunity?
>>> or let me know if discussion is required
>>>
>>> My do not have eyesight/bandwidth to open another tab at the moment
>>> I appreciate the plan is starting to take shape :-)
>>>
>>> Thanks
>>> PDM
>>>
>>>
>>> On Thu, Jul 23, 2020 at 7:30 AM Paola Di Maio <paola.dimaio@gmail.com>
>>> wrote:
>>>
>>>> Carl
>>>>
>>>> Thanks for the updates
>>>>
>>>> As discussed in previous emails, and also during a call, I d like to
>>>> see the AI strategist role to include:
>>>> - ensure KR is adequate to support the logical and functional integrity
>>>> of the system
>>>> - KR is applied to avert the risks of systemic bias
>>>>
>>>> Would you be able to include these, or do I have do to something?
>>>>
>>>> Thank you
>>>> PDM
>>>>
>>>> On Wed, Jul 22, 2020 at 10:24 PM carl mattocks <carlmattocks@gmail.com>
>>>> wrote:
>>>>
>>>>> Please peruse our work-in-progress AI KR Strategist core plan. Your
>>>>> comments will be used to build out a version that has measurable  KPIs
>>>>> for all Objectives .
>>>>>
>>>>> Currently it contains Goals & Objectives and a set of KPIs focused on
>>>>> algorithmic learning within context  :
>>>>>
>>>>>    - For all AI systems to have clearly and transparently documented
>>>>>    goals and performance data showing that they are being achieved.
>>>>>    - The mission of an AI Strategist is to define the purpose and
>>>>>    goals of AI systems, as well as the KPIs by which we can determine if the
>>>>>    system is meeting its goals.
>>>>>
>>>>>
>>>>> Here is a StratML compliant report .. a custom WORD report is attached
>>>>>
>>>>> https://www.stratnavapp.com/Projects/861566c8-e9be-4642-b52f-f673fa499f4e/HTMLReport
>>>>>
>>>>>
>>>>> enjoy
>>>>>
>>>>> Carl Mattocks
>>>>>  co-chair AIKR CG
>>>>> It was a pleasure to clarify
>>>>>
>>>>