Difference between revisions of "ChatGPT4-Questions/User:Darwin2049/Overview"
Jump to navigation
Jump to search
Darwin2049 (talk | contribs) |
Darwin2049 (talk | contribs) |
||
| (31 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
2023.11.09 - PAGE IS NOW ALL SCREWED UP....<BR | <!-- 2023.11.09 - PAGE IS NOW ALL SCREWED UP....<BR> | ||
FIRST ORDER OF BUSINESS FOR TOMORROW IS TO PUT IT BACK IN ORDER<BR | FIRST ORDER OF BUSINESS FOR TOMORROW IS TO PUT IT BACK IN ORDER<BR> | ||
AND FIX UP ANY OTHER DISCREPANCIES; SHOULD BE: (AND EXPLAIN WHY)<BR | AND FIX UP ANY OTHER DISCREPANCIES; SHOULD BE: (AND EXPLAIN WHY)<BR> | ||
IMPRESSIONS<BR | IMPRESSIONS<BR> | ||
OPERATIONS<BR | OPERATIONS<BR> | ||
CAVEATS<BR | CAVEATS<BR> | ||
RISKS<BR | RISKS<BR> | ||
INTERMEDIATE OBSERVATIONS/CONCLUSIONS<BR | INTERMEDIATE OBSERVATIONS/CONCLUSIONS<BR> | ||
PHASE SHIFT<BR | PHASE SHIFT<BR> | ||
SPECULATION<BR | SPECULATION<BR> | ||
CONCLUSION<BR | CONCLUSION<BR> | ||
--> | |||
<!-- | |||
2023.12.25: wow! major step forward; quantum computing - very serious thicket to wade through; very difficult to try to summarize without sufficient | |||
background in physics and mathematics; | |||
--> | |||
[[File:OPENAI.png|left|220px]] | [[File:OPENAI.png|left|220px]] | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">OpenAI - ChatGPT4.</SPAN>''''' <BR | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">OpenAI - ChatGPT4.</SPAN>''''' </BR> | ||
In what follows we attempt to address several basic questions about the onrushing progress with the current focus of artificial intelligence. There are several competing actors in this space. These include OpenAI, DeepMind, Anthropic, and Cohere. A number of other competitors are active in the artificial intelligence market place. But for purposes of brevity and because of the overlap we will limit focus on ChatGPT4 (CG4). Further, we focus on several salient questions that that raise questions of safety, risk and prospects. <BR | In what follows we attempt to address several basic questions about the onrushing progress with the current focus of artificial intelligence. There are several competing actors in this space. These include OpenAI, DeepMind, Anthropic, and Cohere. A number of other competitors are active in the artificial intelligence market place. But for purposes of brevity and because of the overlap we will limit focus on ChatGPT4 (CG4). Further, we focus on several salient questions that that raise questions of safety, risk and prospects. <BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Responses.</SPAN>''''' The following links connect to responses to their respective questions:<BR | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Responses.</SPAN>''''' The following links connect to responses to their respective questions:<BR> | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Conclusions_Synthesis Interfacing/Accessibility-Conformability - Synthesis.]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how will different groups interact with, respond to and be affected by it; might access modalities available to one group have positive or negative implications for other groups; </SPAN> | *'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Conclusions_Synthesis Interfacing/Accessibility-Conformability - Synthesis.]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how will different groups interact with, respond to and be affected by it; might access modalities available to one group have positive or negative implications for other groups; </SPAN> | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Darwin2049/chatgpt4_omega_political Political/Competitive - Synthesis.]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how might different groups or actors gain or lose relative advantage; also, how might it be used as a tool of control;</SPAN> | *'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Darwin2049/chatgpt4_omega_political Political/Competitive - Synthesis.]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how might different groups or actors gain or lose relative advantage; also, how might it be used as a tool of control;</SPAN> | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Darwin2049/chatgpt4_omega_evolutionary Evolutionary/Stratification - Synthesis.]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> might new classifications of social categories emerge; were phenotypical bifurcations to emerge would or how would the manifest themselves;</SPAN><BR /> | *'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Darwin2049/chatgpt4_omega_evolutionary Evolutionary/Stratification - Synthesis.]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> might new classifications of social categories emerge; were phenotypical bifurcations to emerge would or how would the manifest themselves;</SPAN><BR /> | ||
* '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Https://arguably.io/User:Darwin2049/chatgpt4_omega_epistemological Epistemological - Synthesis]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how to reconcile ethical issues within a society, between societies; more specifically, might it provide solutions or results that are acceptable to the one group but unacceptable to the other group;</SPAN><BR | * '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Https://arguably.io/User:Darwin2049/chatgpt4_omega_epistemological Epistemological - Synthesis]</SPAN>''''' <Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how to reconcile ethical issues within a society, between societies; more specifically, might it provide solutions or results that are acceptable to the one group but unacceptable to the other group;</SPAN><BR> | ||
<BR /> | <BR /> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Summary.</SPAN>''''' The '''''<Span Style="COLOR: | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Summary.</SPAN>''''' The '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">links above</SPAN>''''' point to this observer's assessment on how to answer the questions originally; <BR> | ||
The '''''<Span Style="COLOR: | The '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">links below</SPAN>''''' point to the steps that underpin the answers above. Getting to these answers involved: <BR> | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">reify the questions: "in other words...":</SPAN>''''' | *'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">reify the questions: "in other words...":</SPAN>''''' reframe and contextualize the questions; | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">sentiment: "so... people are saying...":</SPAN>''''' this meant having a quick glance at the rapidly growing body of reviews and opinions; the resulting observations showed that there is a broad range of sentiment to enthusiasm to panic; | *'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">sentiment: "so... people are saying...":</SPAN>''''' this meant having a quick glance at the rapidly growing body of reviews and opinions; the resulting observations showed that there is a broad range of sentiment to enthusiasm to panic; | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">how does it work: "just push that button!":</SPAN>''''' without getting too deep into the weeds, asking how an LLM works; early uses of the most popular system CG4 showed intrinsic shortcomings that obliged attention; other Deep Learning systems were examined superficially but coverage of their relative features is deferred for another round of analysis; | *'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">how does it work: "just push that button!":</SPAN>''''' without getting too deep into the weeds, asking how an LLM works; early uses of the most popular system CG4 showed intrinsic shortcomings that obliged attention; other Deep Learning systems were examined superficially but coverage of their relative features is deferred for another round of analysis; | ||
| Line 30: | Line 33: | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">caveats: "however..." :</SPAN>''''' given what is now known about these new capabilities it appeared to be useful to offer some qualifying caveats; these can be expanded upon, dismissed or modified as needed; | *'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">caveats: "however..." :</SPAN>''''' given what is now known about these new capabilities it appeared to be useful to offer some qualifying caveats; these can be expanded upon, dismissed or modified as needed; | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">o.k. so far then: "so what if...":</SPAN>''''' have been put forward as a way to approach the more speculative and theoretical prospects of where events may go; | *'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">o.k. so far then: "so what if...":</SPAN>''''' have been put forward as a way to approach the more speculative and theoretical prospects of where events may go; | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">conclusions - " | *'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">conclusions - "so... this must mean that...":</SPAN>''''' what do we know now; can we be sure, things are moving fast? can we just say that this is what we believe so far?;<BR> | ||
<!-- | <!-- | ||
Attempting to address these issues suggested that a sound basis for forward motion meant to reify them in terms of the implications that they present. The way that they are presented suggest that this new LLM technology entails risks to various constituencies. In order to characterize these risk meant that they needed to be identified. The nature of risk involve the potential for payoff. Which then suggested answering them in terms of game theory. Therefore how to view these various constituencies in terms of zero/non-zero sum outcomes. In order to gain leverage on that question meant getting an overview of sentiment. The sentiment used in this discourse involved a snap review of what some of the most informed observers were saying. What followed from this assessment was an evaluation of just what an LLM is and how it compares to other forms of Deep Learning systems. | Attempting to address these issues suggested that a sound basis for forward motion meant to reify them in terms of the implications that they present. The way that they are presented suggest that this new LLM technology entails risks to various constituencies. In order to characterize these risk meant that they needed to be identified. The nature of risk involve the potential for payoff. Which then suggested answering them in terms of game theory. Therefore how to view these various constituencies in terms of zero/non-zero sum outcomes. In order to gain leverage on that question meant getting an overview of sentiment. The sentiment used in this discourse involved a snap review of what some of the most informed observers were saying. What followed from this assessment was an evaluation of just what an LLM is and how it compares to other forms of Deep Learning systems. | ||
<BR | <BR> | ||
--> | --> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/chatgpt4_impressions Impressions]</SPAN>''''' Sentiments that fell into a few categories: | '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/chatgpt4_impressions Impressions]</SPAN>''''' Sentiments that fell into a few categories: | ||
'''''positive (this is going to be great!)''''' , '''''cautious (not so fast!...) ''''' (or cautiously worried) or '''''alarmed (it might kill us all!).''''' Geopolitical observers expressed | '''''positive (this is going to be great!)''''', '''''cautious (not so fast!...) ''''' (or cautiously worried) or '''''alarmed (it might kill us all!).''''' Geopolitical observers expressed existential threats from rivals - ''''' pedal to the metal!'''''; <BR> | ||
<!-- | <!-- | ||
'''''<Span Style="COLOR:WHITE;BACKGROUND:BLUE">Impressions.</SPAN>''''' Addressing these question indicated that making a short survey of reporting with the intent to gauge the reported sentiment might offer some insight. The results suggested four major categories. These sentiments included ('''''positives''''' favor continued advances '''''worried''''' cautious, worried; '''''alarmed''''' petitioned for government intervention; '''''battle stations!''''' being overtaken by a rival is unacceptable);<BR | '''''<Span Style="COLOR:WHITE;BACKGROUND:BLUE">Impressions.</SPAN>''''' Addressing these question indicated that making a short survey of reporting with the intent to gauge the reported sentiment might offer some insight. The results suggested four major categories. These sentiments included ('''''positives''''' favor continued advances '''''worried''''' cautious, worried; '''''alarmed''''' petitioned for government intervention; '''''battle stations!''''' being overtaken by a rival is unacceptable);<BR> --> | ||
<!-- The way that these sentiments were categorized in terms of risk involved describing the kind of risk that each seemed to be either stating explicitly or implying.<BR> | |||
<!-- The way that these sentiments were categorized in terms of risk involved describing the kind of risk that each seemed to be either stating explicitly or implying.<BR | |||
An analysis of sentiment and impressions could easily become a research effort in and of itself. That did not appear to be the primary objective of the examination. Therefore only a small sampling of well publicized reports have been included. Possibly a more detailed approach might be indicated at a later time. | An analysis of sentiment and impressions could easily become a research effort in and of itself. That did not appear to be the primary objective of the examination. Therefore only a small sampling of well publicized reports have been included. Possibly a more detailed approach might be indicated at a later time. | ||
--> | |||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4_Operations Operations]</SPAN>''''', Understanding the CG4 internal mechanisms might offer some insight into how it does what it does. And therefore by extension how one group might gain or lose advantage. The result is that several variants of the Deep Learning approach came to light but the Large Language Model (LLM) seemed to be the preferred point of entre'. This was because it has shown itself to be remarkably versatile in its range of applicability. <BR> | |||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Risks Risks.]</SPAN>''''' Three categories emerged: systemic, malicious and theoretical. In each case our observation is that this new technology is inherently dual use.<BR> That this technology does show itself to be dual use led to the intimation that a pause for some considerations was in order before proceeding. They led to the intermediate synthesis that can be found next. | |||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Risks Risks.]''''' | |||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4_Intermediate_Considerations Intermediate Synthesis]</SPAN>''''' Based upon what we have observed | '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4_Intermediate_Considerations Intermediate Synthesis]</SPAN>''''' Based upon what we have observed | ||
our deliberations suggested that we make more explicit what we think and feel as well as offering some caveats for further consideration.<BR | our deliberations suggested that we make more explicit what we think and feel as well as offering some caveats for further consideration.<BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/chatgpt4_caveats Caveats]''''' | '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/chatgpt4_caveats Caveats]</SPAN>''''' Our analysis to this point has suggested that several crucial factors be acknowledged. These include such observations that Deep Learning technology results are dual use. They can be used to further facilitate social, economic and political well being. But they can also be used for malicious purposes that can not yet be imagined.<BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Darwin2049/ChatGPT4/PhaseShift Phase Shift.]'''''< | '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Darwin2049/ChatGPT4/PhaseShift Phase Shift.]</SPAN>''''' Quantum computing is expected to momentarily make its debut by or before the end of the year. IBM will announce availability of its '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">IBM Quantum System Two</SPAN>''''' (EOY 2023). This recently available systems is based on a scalable ensemble of '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">432 Q-Bit Osprey</SPAN>''''' devices. Migrating Deep Learning systems to a quantum computing environment will result in a quantum computing capability of over 1000 qubits and represents a '''''<Span Style="COLOR:BLUE;BACKGROUND:SILVER">before/after</SPAN>''''' event. <BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Theoretical Theoretical]</SPAN>''''' These are speculative. That being so, they attempt to avoid going beyond the bounds of the possible. <BR> | |||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Conclusions_Synthesis Conclusions Synthesis]''''' | '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/ChatGPT4/Conclusions_Synthesis Conclusions Synthesis]</SPAN>''''' Finally we try to arrive at what supports our position regarding how the question groups were answered. These were the questions on interfaces, political, evolutionary and epistemology. <BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">Note & References''''' | '''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">Note & References</SPAN>''''' The notes and references that follow are intended to provide further support to the theses promoted in this effort. | ||
<!-- *********************************************************************************************************** --> | <!-- *********************************************************************************************************** --> | ||
<!-- *********************************************************************************************************** --> | <!-- *********************************************************************************************************** --> | ||
| Line 105: | Line 104: | ||
<!-- | <!-- | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Https://arguably.io/User:Darwin2049/chatgpt4_omega_epistemological Epistemological/Ethical Relativism - Synthesis]</SPAN>'''''<Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how to reconcile ethical issues within a society, between societies; more specifically, might it provide solutions or results that are acceptable to the one group but unacceptable to the other group;<Span Style="COLOR:WHITE; BACKGROUND:BLUE"><BR | *'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/Https://arguably.io/User:Darwin2049/chatgpt4_omega_epistemological Epistemological/Ethical Relativism - Synthesis]</SPAN>'''''<Span Style="COLOR:WHITE; BACKGROUND:TEAL"> how to reconcile ethical issues within a society, between societies; more specifically, might it provide solutions or results that are acceptable to the one group but unacceptable to the other group;<Span Style="COLOR:WHITE; BACKGROUND:BLUE"><BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Synthesis.</SPAN>''''' <BR | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Synthesis.</SPAN>''''' <BR> These four question groups (''<Span Style="COLOR:WHITE; BACKGROUND:BLUE">interface, political, evolution</SPAN> and <Span Style="COLOR:WHITE; BACKGROUND:BLUE">epistemological</SPAN>'') form the core of what follows. Attempting to respond to these questions obliged the development of a review and response plan. What followed is a set of steps that have attempted to shed light on each of these areas; over the course of this examination the reality has been that addressing these four groups was not a straightforward one to one mapping; | ||
Instead, addressing them has meant gaining focus on informed perceptions by knowledge able observers; from there it was possible to proceed to try to answer the implicit questions relating to risk; in order to understand risk entailed getting a better picture of what underpins the CG4 technology and by corollary its peer architectures; by way of identifying and articulating the various risk categories only a short step was indicated to ask the question of "what happens now?"; this is presented in a speculative step into a near future when new tools and processes have become accepted | Instead, addressing them has meant gaining focus on informed perceptions by knowledge able observers; from there it was possible to proceed to try to answer the implicit questions relating to risk; in order to understand risk entailed getting a better picture of what underpins the CG4 technology and by corollary its peer architectures; by way of identifying and articulating the various risk categories only a short step was indicated to ask the question of "what happens now?"; this is presented in a speculative step into a near future when new tools and processes have become accepted | ||
Therefore this examination proceeds along the following lines: | Therefore this examination proceeds along the following lines: | ||
| Line 117: | Line 116: | ||
By of proceeding, we: | By of proceeding, we: | ||
*'''''spotlight the main AI approaches:''''' show several comparable technologies (''generative, transformers, bidirectional, bidirectional encoders, generative adversarial, convolutional, recursive, large language models.'');<BR | *'''''spotlight the main AI approaches:''''' show several comparable technologies (''generative, transformers, bidirectional, bidirectional encoders, generative adversarial, convolutional, recursive, large language models.'');<BR> | ||
*'''''focus risk questions:''''' the four question groups mentioned above implicitly suggest risk; the following analysis offers several examples that suggest how CG4 class systems will extend the realities of risk; ''systemic, malicious and theoretical''); in the case of theoretical we would add that caveats apply when reviewing this group because of the inherent dual use capabilities that CG4 offers; | *'''''focus risk questions:''''' the four question groups mentioned above implicitly suggest risk; the following analysis offers several examples that suggest how CG4 class systems will extend the realities of risk; ''systemic, malicious and theoretical''); in the case of theoretical we would add that caveats apply when reviewing this group because of the inherent dual use capabilities that CG4 offers; | ||
| Line 133: | Line 132: | ||
<!-- ************************************************************************************* --> | <!-- ************************************************************************************* --> | ||
<!-- | <!-- | ||
*'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/chatgpt4_caveats Caveats.]</SPAN>''''' <BR | *'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">[https://arguably.io/User:Darwin2049/chatgpt4_caveats Caveats.]</SPAN>''''' <BR>After assessing the range of sentiment from very well informed observers and experts our conclusion is that a summary of caveats and qualifications might help in constraining and focusing the discussion going forward. | ||
* '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">An inflection point</SPAN>''''' either has or momentarily will be or has been reached; | * '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">An inflection point</SPAN>''''' either has or momentarily will be or has been reached; | ||
* '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">mitigating surprise</SPAN>''''' the concerned community would be well advised to regularly monitor the archive for papers so as to mitigate the gain early warning; | * '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">mitigating surprise</SPAN>''''' the concerned community would be well advised to regularly monitor the archive for papers so as to mitigate the gain early warning; | ||
| Line 171: | Line 170: | ||
** uses a SAAS model; like Google Search, Youtube or Morningstar Financial; | ** uses a SAAS model; like Google Search, Youtube or Morningstar Financial; | ||
--> | --> | ||
<BR | <BR>'''''<Span Style="COLOR:BLUE; BACKGROUND:YELLOW">Back to root page: [https://arguably.io/ChatGPT4-Questions ChatGPT4-Questions]</SPAN>''''' | ||
<!-- ************************************************************************************* --> | <!-- ************************************************************************************* --> | ||
<!-- ************************************************************************************* --> | <!-- ************************************************************************************* --> | ||
<!-- | <!-- | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Original Questions.</SPAN>''''' Formulating a response to these questions took the approach of scanning whatever news reports seemed to make mention of the topic; from there look for sentiment patterns or groupings;<BR | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Original Questions.</SPAN>''''' Formulating a response to these questions took the approach of scanning whatever news reports seemed to make mention of the topic; from there look for sentiment patterns or groupings;</BR> as this process progressed | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Sentiment.</SPAN>''''' In order to maintain focus and forward momentum sentiment has been expressed as voices arguing in favor of further development. Other notable voices have urged caution going forward. Not least among them were such notable figures as Elon Musk. Voices expressing the same sentiment but with greater force suggested that all efforts going forward with this research should be paused for a limited amount of time. Then there were voices that expressed grave concern that this represents an existential all-of-national-resources imperative to go pedal to the metal. The risk to slow down or pause while avowed rivals are pressing ahead was just too great. In short the main camps that seem to have coalesced include: '''''press on, proceed with caution, severe storm warnings ahead, damn the torpedoes full speed ahead.'''''<BR | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Sentiment.</SPAN>''''' In order to maintain focus and forward momentum sentiment has been expressed as voices arguing in favor of further development. Other notable voices have urged caution going forward. Not least among them were such notable figures as Elon Musk. Voices expressing the same sentiment but with greater force suggested that all efforts going forward with this research should be paused for a limited amount of time. Then there were voices that expressed grave concern that this represents an existential all-of-national-resources imperative to go pedal to the metal. The risk to slow down or pause while avowed rivals are pressing ahead was just too great. In short the main camps that seem to have coalesced include: '''''press on, proceed with caution, severe storm warnings ahead, damn the torpedoes full speed ahead.'''''<BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Literature Review.</SPAN>''''' Therefore the path of progress proceeded along the lines of: what are informed people saying about this new capability was foretelling. Reporting seemed to be increasing by the day. This necessitated limiting the choices of what to view and include since trying to gather it all could become a research project in and of itself. | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Literature Review.</SPAN>''''' Therefore the path of progress proceeded along the lines of: what are informed people saying about this new capability was foretelling. Reporting seemed to be increasing by the day. This necessitated limiting the choices of what to view and include since trying to gather it all could become a research project in and of itself. | ||
Based upon these sentiment it should come as no surprise that the risks are real and pressing. Anyone viewing a real time world map of malicious attacks shows a planet awash in twenty-four seven activity.<BR | Based upon these sentiment it should come as no surprise that the risks are real and pressing. Anyone viewing a real time world map of malicious attacks shows a planet awash in twenty-four seven activity.<BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Model Variations.</SPAN>''''' By this point in the examination it became evident that putting a spotlight the underlying technology might help. This revealed that there were several variations on the underlying approach of Deep Learning systems. Prominent ones included: | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Model Variations.</SPAN>''''' By this point in the examination it became evident that putting a spotlight the underlying technology might help. This revealed that there were several variations on the underlying approach of Deep Learning systems. Prominent ones included: | ||
'''''generative, transformers, bidirectional, bidirectional encoders, generative adversarial, convolutional, recursive, large language models.'''''<BR | '''''generative, transformers, bidirectional, bidirectional encoders, generative adversarial, convolutional, recursive, large language models.'''''<BR> | ||
We briefly touch on each of these but the main focus has been on the LLM approach that uses a Pretrained approach, hence the name GPT... generative pretrained transformers.<BR | We briefly touch on each of these but the main focus has been on the LLM approach that uses a Pretrained approach, hence the name GPT... generative pretrained transformers.<BR> | ||
'''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Risks.</SPAN>''''' Therefore the immediate follow on question became given that this is a major advance in a known technology how might it be incorporated, the answer to that rhetorical question fell right out into plain view as being how might this new technology pose risks, and what kinds of risks. These immediately fell into three main categories... i.e. risks that were organic (they arose innately as a result of their existence), malicious (actors with malevolent intent) and theoretical (possible new developments that are possible or newly enabled).<BR | '''''<Span Style="COLOR:BLUE; BACKGROUND:SILVER">Risks.</SPAN>''''' Therefore the immediate follow on question became given that this is a major advance in a known technology how might it be incorporated, the answer to that rhetorical question fell right out into plain view as being how might this new technology pose risks, and what kinds of risks. These immediately fell into three main categories... i.e. risks that were organic (they arose innately as a result of their existence), malicious (actors with malevolent intent) and theoretical (possible new developments that are possible or newly enabled).<BR> | ||
--> | --> | ||
Latest revision as of 02:12, 26 December 2023
OpenAI - ChatGPT4.
In what follows we attempt to address several basic questions about the onrushing progress with the current focus of artificial intelligence. There are several competing actors in this space. These include OpenAI, DeepMind, Anthropic, and Cohere. A number of other competitors are active in the artificial intelligence market place. But for purposes of brevity and because of the overlap we will limit focus on ChatGPT4 (CG4). Further, we focus on several salient questions that that raise questions of safety, risk and prospects.
Responses. The following links connect to responses to their respective questions:
- Interfacing/Accessibility-Conformability - Synthesis. how will different groups interact with, respond to and be affected by it; might access modalities available to one group have positive or negative implications for other groups;
- Political/Competitive - Synthesis. how might different groups or actors gain or lose relative advantage; also, how might it be used as a tool of control;
- Evolutionary/Stratification - Synthesis. might new classifications of social categories emerge; were phenotypical bifurcations to emerge would or how would the manifest themselves;
- Epistemological - Synthesis how to reconcile ethical issues within a society, between societies; more specifically, might it provide solutions or results that are acceptable to the one group but unacceptable to the other group;
Summary. The links above point to this observer's assessment on how to answer the questions originally;
The links below point to the steps that underpin the answers above. Getting to these answers involved:
- reify the questions: "in other words...": reframe and contextualize the questions;
- sentiment: "so... people are saying...": this meant having a quick glance at the rapidly growing body of reviews and opinions; the resulting observations showed that there is a broad range of sentiment to enthusiasm to panic;
- how does it work: "just push that button!": without getting too deep into the weeds, asking how an LLM works; early uses of the most popular system CG4 showed intrinsic shortcomings that obliged attention; other Deep Learning systems were examined superficially but coverage of their relative features is deferred for another round of analysis;
- risks: "this is great/horrible news!": almost all new technologies involve risk of some sort; these might inherent or organic simply because of their existence or intentionally malicious; past experience provides a framework to speculate about what might come knowing what has been done before; some examples are provided to offer context for further discussion;
- caveats: "however..." : given what is now known about these new capabilities it appeared to be useful to offer some qualifying caveats; these can be expanded upon, dismissed or modified as needed;
- o.k. so far then: "so what if...": have been put forward as a way to approach the more speculative and theoretical prospects of where events may go;
- conclusions - "so... this must mean that...": what do we know now; can we be sure, things are moving fast? can we just say that this is what we believe so far?;
Impressions Sentiments that fell into a few categories:
positive (this is going to be great!), cautious (not so fast!...) (or cautiously worried) or alarmed (it might kill us all!). Geopolitical observers expressed existential threats from rivals - pedal to the metal!;
Operations, Understanding the CG4 internal mechanisms might offer some insight into how it does what it does. And therefore by extension how one group might gain or lose advantage. The result is that several variants of the Deep Learning approach came to light but the Large Language Model (LLM) seemed to be the preferred point of entre'. This was because it has shown itself to be remarkably versatile in its range of applicability.
Risks. Three categories emerged: systemic, malicious and theoretical. In each case our observation is that this new technology is inherently dual use.
That this technology does show itself to be dual use led to the intimation that a pause for some considerations was in order before proceeding. They led to the intermediate synthesis that can be found next.
Intermediate Synthesis Based upon what we have observed
our deliberations suggested that we make more explicit what we think and feel as well as offering some caveats for further consideration.
Caveats Our analysis to this point has suggested that several crucial factors be acknowledged. These include such observations that Deep Learning technology results are dual use. They can be used to further facilitate social, economic and political well being. But they can also be used for malicious purposes that can not yet be imagined.
Phase Shift. Quantum computing is expected to momentarily make its debut by or before the end of the year. IBM will announce availability of its IBM Quantum System Two (EOY 2023). This recently available systems is based on a scalable ensemble of 432 Q-Bit Osprey devices. Migrating Deep Learning systems to a quantum computing environment will result in a quantum computing capability of over 1000 qubits and represents a before/after event.
Theoretical These are speculative. That being so, they attempt to avoid going beyond the bounds of the possible.
Conclusions Synthesis Finally we try to arrive at what supports our position regarding how the question groups were answered. These were the questions on interfaces, political, evolutionary and epistemology.
Note & References The notes and references that follow are intended to provide further support to the theses promoted in this effort.
Back to root page: ChatGPT4-Questions