Hallo.

Fragen? Ein erstes Kennenlernen? Vielleicht hast Du schon konkrete Pläne? Du kannst Dich jederzeit unverbindlich an uns wenden!

Telefon: 0761 - 76 99 34 20
Email: info@maexware-solutions.de

über Kontaktformular:

Mit Absenden des Kontaktformulars erklärst Du Dich damit einverstanden, dass Deine Daten zur Bearbeitung Deines Anliegens verwendet werden. Weitere Informationen kannst Du der Datenschutzerklärung entnehmen.

Erste Versuche mit ChatGPT - Teil 3

Erste Versuche mit ChatGPT - Teil 3

- Monib Asad Masud

Im ersten und zweiten Artikel haben wir schon ein paar Learnings bei der Zusammenarbeit mit ChatGPT machen können.

Für diesen dritten Artikel-Teil wollen wir uns damit beschäftigen, was ChatGPT erzeugt, wenn ihm kein Kontext vorliegt, auf dem es aufbauen kann. Jede Aufgabenstellung erzeugt für sich einen Kontext, aber wir wollen das “Wissen” des trainierten Sprachmodells auffordern zu zeigen was es kann. Mal schauen, ob wir dafür die richtigen, möglichst kurzen, Fragestellungen, bzw. “Formulierungen” finden.

Wir schreiben ChatGPT in einer frischen Konversation
“We need a template to protocol.”

Und ChatGPT, blitzschnell:
“Sure! Here's a template for a protocol:

[Title of the Protocol]

[Objective] Clearly state the objective of the protocol, which should be a concise and specific description of what the protocol aims to accomplish.

[Background] Provide a brief overview of the background or context of the protocol, including any relevant literature or existing methods that influenced the development of the protocol.

[Materials] List all the materials required to perform the protocol, including any specialized equipment, reagents, or software.

[Methods] Describe the detailed steps necessary to perform the protocol, including any safety precautions, data collection procedures, or data analysis methods. It should be written in a clear and concise manner to ensure reproducibility.

[Results] Specify what results are expected to be obtained by the protocol. Include any expected outcome, data interpretation or statistical analysis.

[Discussion] Discuss any limitations, alternative methods or any observations made during the performance of the protocol. Also, mention any future directions for improvement or modification of the protocol.

[Conclusion] Summarize the overall purpose and significance of the protocol, including any implications for future research or practice.

[References] List all the references cited in the protocol.

[Appendix] Include any supplementary materials that may be relevant to the protocol, such as diagrams, flowcharts, tables, or additional data.

I hope this template is helpful!”

Das “fühlte” sich jetzt sehr sauber an. Frage. Antwort. Fertig. ChatGPT weiß nichts über unsere Absichten, unsere Vorstellungen oder wohin das ganze führen soll. Es sieht in dieser frischen Konversation nur diesen einen Satz, unsere “letzte” Frage, und das Sprachmodell erzeugt etwas, was es durch sein Training am geeignetsten hält, auf diesen Satz zu reagieren.

Diesen Abgleich von Frage zu Antwort, können wir auch eine Iteration nennen. Dies ist das, was ChatGPT tut, egal wie lange eine Konversation andauert oder wie viele Iterationen zuvor geschahen. Selbst wenn wir uns dazu entscheiden eine Konversation als Ganzes zu betrachten, ChatGPT tut dies “so” nicht. ChatGPT geht immer von der “jetzigen” Iteration aus. Dass ChatGPT “Zugriff” auf die vorherigen Iterationen hat, wird “nur” dazu genutzt die “jetzige” Iteration “besser” zu beantworten. Das “besser” ist dabei mit Vorsicht zu genießen, wie wir gelernt haben, denn je mehr wir als Mensch in die Konversation einbringen, umso höher ist die Wahrscheinlichkeit, dass ChatGPT Aspekte in seine Antworten mit einbezieht, die nichts mit unserem gewünschten Ergebnis zu tun haben. Wir wissen, welche Teile wir von unseren Interationen berücksichtigt haben wollen, ChatGPT nicht.

Haben wir nun eigentlich durch diese eine Iteration, das war wir haben wollen, gemessen an unseren anderen Artikel-Teilen? Nein. Aber was wir beibehalten wollen ist diese “Sauberkeit”, mit der die erste Iteration gekennzeichnet war. Uns fallen dazu zwei Ideen ein. Genauso weitermachen in dieser Konversation, oder das Ergebnis nehmen und eine neue Konversation starten.

Wir versuchen so “sauber” weiter zu machen in der jetzigen Konversation. Wir schreiben:

“The template to protocol should reflect software development, requirement management, costs, names of participants, timelines.”

Und ChatGPT, wieder ohne uns lange warten zu lassen:

“Certainly! Here's an updated protocol template that includes sections for software development, requirement management, costs, participant names, and timelines:

[Title of the Protocol]

[Objective] Clearly state the objective of the protocol, which should be a concise and specific description of what the protocol aims to accomplish.

[Background] Provide a brief overview of the background or context of the protocol, including any relevant literature or existing methods that influenced the development of the protocol.

[Materials] List all the materials required to perform the protocol, including any specialized equipment, reagents, or software.

[Software Development] Describe the software development methodology that will be used, such as Agile, Waterfall, or Scrum, and provide a timeline for the development process.

[Requirement Management] Specify the tools and processes that will be used for managing the requirements of the software, such as a requirements traceability matrix or a requirements management tool.

[Costs] Provide an estimate of the costs associated with the development of the software, including any hardware or software licenses, development costs, and maintenance costs.

[Participant Names] List the names and roles of all participants involved in the development of the software, such as developers, testers, and project managers.

[Timelines] Provide a detailed timeline for the development of the software, including milestones, deliverables, and deadlines.

[Methods] Describe the detailed steps necessary to perform the protocol, including any safety precautions, data collection procedures, or data analysis methods. It should be written in a clear and concise manner to ensure reproducibility.

[Results] Specify what results are expected to be obtained by the protocol. Include any expected outcome, data interpretation or statistical analysis.

[Discussion] Discuss any limitations, alternative methods or any observations made during the performance of the protocol. Also, mention any future directions for improvement or modification of the protocol.

[Conclusion] Summarize the overall purpose and significance of the protocol, including any implications for future research or practice.

[References] List all the references cited in the protocol.

[Appendix] Include any supplementary materials that may be relevant to the protocol, such as diagrams, flowcharts, tables, or additional data.

I hope this template is more suitable for your needs!”

Wenn wir jetzt die erste und zweite Iteration betrachten, haben wir keine wahnsinnigen Veränderungen/Fortschritte gemacht, aber alles ist immer noch sehr sauber - und ChatGPT hat genau das getan, was wir für diese Iteration wollten. Versuchen wir eine weitere Iteration und bleiben wir so “sauber” wie möglich. Wir wissen, wo wir hin wollen, aber wir dürfen nicht “gierig” werden.

Wir schreiben ChatGPT:

“The Template to protocol should contain proposed sub-steps or subpoints to every major protocol point.”

Und ChatGPT erwiedert:

“Certainly! Here's a revised template for a protocol that includes proposed sub-steps or subpoints for every major protocol point:

[Title of the Protocol]

[Objective] Clearly state the objective of the protocol, which should be a concise and specific description of what the protocol aims to accomplish.

[Background] Provide a brief overview of the background or context of the protocol, including any relevant literature or existing methods that influenced the development of the protocol.

[Materials] List all the materials required to perform the protocol, including any specialized equipment, reagents, or software.

[Software Development] Describe the software development methodology that will be used, such as Agile, Waterfall, or Scrum, and provide a timeline for the development process.

Sub-points:

  • Define the scope of the software development project
  • Identify the key stakeholders involved in the software development process
  • Determine the development team's roles and responsibilities
  • Develop a project plan and timeline that includes key milestones and deadlines
  • Conduct regular project status meetings to track progress and identify any issues or risks


[Requirement Management] Specify the tools and processes that will be used for managing the requirements of the software, such as a requirements traceability matrix or a requirements management tool.

Sub-points:

  • Develop a requirements gathering plan
  • Identify the key stakeholders who will provide input on the software requirements
  • Define the software requirements and prioritize them based on importance and feasibility
  • Document the software requirements and create a traceability matrix
  • Conduct regular reviews of the requirements to ensure they are complete and accurate


[Costs] Provide an estimate of the costs associated with the development of the software, including any hardware or software licenses, development costs, and maintenance costs.

Sub-points:

  • Identify all hardware and software requirements
  • Estimate the cost of hardware and software licenses
  • Estimate the development costs, including labor and any other resources required
  • Identify any ongoing maintenance costs and estimate the total cost of ownership


[Participant Names] List the names and roles of all participants involved in the development of the software, such as developers, testers, and project managers.

Sub-points:

  • Identify all project stakeholders
  • Define the roles and responsibilities of each team member
  • Create an organizational chart that identifies the reporting structure


[Timelines] Provide a detailed timeline for the development of the software, including milestones, deliverables, and deadlines.

Sub-points:

  • Define the project schedule and identify key milestones
  • Develop a detailed timeline that includes all development phases, such as requirements gathering, design, development, testing, and deployment
  • Create a Gantt chart to visualize the timeline and track progress


[Methods] Describe the detailed steps necessary to perform the protocol, including any safety precautions, data collection procedures, or data analysis methods. It should be written in a clear and concise manner to ensure reproducibility.

Sub-points:

  • Define the data collection process and data analysis methods
  • Identify any safety precautions that need to be taken during the development process
  • Create a detailed step-by-step plan for the development process
  • Develop a testing plan to ensure the software meets all requirements


[Results] Specify what results are expected to be obtained by the protocol. Include any expected outcome, data interpretation or statistical analysis.

Sub-points:

  • Define the criteria for success
  • Identify the expected outcomes of the protocol
  • Develop a plan for interpreting and analyzing the data collected during the development process


[Discussion] Discuss any limitations, alternative methods or any observations made during the performance of the protocol. Also, mention any future directions for improvement or modification of the protocol.

Sub-points:

  • Discuss any limitations or challenges encountered during the development process
  • Identify any alternative methods that could have been used to achieve the objective
  • Discuss potential future directions for improving or modifying the software development process


[Conclusion] Summarize the overall purpose and significance of the protocol, including any implications for future research or practice.

[References] List all the references cited in the protocol.

[Appendix] Include any supplementary materials that may be relevant to the protocol, such as diagrams, flowcharts, tables, or additional data.

I hope this updated template with sub-points for each major protocol point is helpful for your needs!”

Sehr interessant, wie sauber Iterationen sich auch nach der Dritten immer noch anfühlen können. Vor allem aber ist interessant, wo wir jetzt stehen. Vergleichen wir das jetzige Ergebnis mit dem, was wir in den anderen Artikel-Teilen bis zu diesem Punkt erreicht hatten, fällt auf, dass wir viel mehr textlichen Inhalt haben, mehr Beschreibung, mehr “Substanz” auf der sich weiter aufbauen lässt (ohne das wir explizit darauf pochen mussten).

Der Aspekt, dass wir keine weiteren Rahmenbedingungen festgelegt haben, bzw. keine Verweise auf Anwendungen oder sogar unser eigentliches Endziel scheint dabei eine große Rolle zu spielen. Und es macht auch Sinn, wenn man sich klar macht, dass man so Zugriff auf das gesamte trainierte Wissen von ChatGPT bekommt, bzw. wenn man das Spektrum des Zugriffs darauf so wenig limitiert wir möglich.

Der tatsächliche Inhalt mag möglicherweise immer noch nicht dem perfekten Zustand entsprechen, den man sich vorstellt, aber auf diese iterative Weise scheint man, sprichwörtlich Schritt für Schritt, sehr gut zu seinem Ziel zu gelangen und gleichzeitig maximalen Nutzen aus dem Wissen von ChatGPT ziehen zu können.

Bevor wir aber zu unseren nächsten Learnings kommen, müssen wir zumindest noch das ganze in Markup besorgen - wie könnte es auch anders sein, sonst wäre ja die Aufgabe nicht erfüllt - also, tun wir es:

“Provide the Template in Markup, use the following Markup as reference.“

# Heading 1
## Heading 2
### Heading 3

- List item 1
- List item 2
- List item 3

1. Numbered list item 1
2. Numbered list item 2
3. Numbered list item 3

**Bold text**
_Italic text_
~~Strikethrough text~~

[Link text](https://www.example.com)

> Blockquote

Code block:
```
Code goes here
```

Inline code: `Code goes here`

Table:

| Column 1 | Column 2 |
|----------|----------|
| Row 1, Column 1 | Row 1, Column 2 |
| Row 2, Column 1 | Row 2, Column 2 |

Und ChatGPT erzeugt

“Sure! Here's the complete updated protocol template in markup:“

# [Title of the Protocol]

## Objective
Clearly state the objective of the protocol, which should be a concise and specific description of what the protocol aims to accomplish.

## Background
Provide a brief overview of the background or context of the protocol, including any relevant literature or existing methods that influenced the development of the protocol.

## Materials
List all the materials required to perform the protocol, including any specialized equipment, reagents, or software.

## Software Development
Describe the software development methodology that will be used, such as Agile, Waterfall, or Scrum, and provide a timeline for the development process.

### Sub-points:
- Define the scope of the software development project
- Identify the key stakeholders involved in the software development process
- Determine the development team's roles and responsibilities
- Develop a project plan and timeline that includes key milestones and deadlines
- Conduct regular project status meetings to track progress and identify any issues or risks

## Requirement Management
Specify the tools and processes that will be used for managing the requirements of the software, such as a requirements traceability matrix or a requirements management tool.

### Sub-points:
- Develop a requirements gathering plan
- Identify the key stakeholders who will provide input on the software requirements
- Define the software requirements and prioritize them based on importance and feasibility
- Document the software requirements and create a traceability matrix
- Conduct regular reviews of the requirements to ensure they are complete and accurate

## Costs
Provide an estimate of the costs associated with the development of the software, including any hardware or software licenses, development costs, and maintenance costs.

### Sub-points:
- Identify all hardware and software requirements
- Estimate the cost of hardware and software licenses
- Estimate the development costs, including labor and any other resources required
- Identify any ongoing maintenance costs and estimate the total cost of ownership

## Participant Names
List the names and roles of all participants involved in the development of the software, such as developers, testers, and project managers.

### Sub-points:
- Identify all project stakeholders
- Define the roles and responsibilities of each team member
- Create an organizational chart that identifies the reporting structure

## Timelines
Provide a detailed timeline for the development of the software, including milestones, deliverables, and deadlines.

### Sub-points:
- Define the project schedule and identify key milestones
- Develop a detailed timeline that includes all development phases, such as requirements gathering, design, development, testing, and deployment
- Create a Gantt chart to visualize the timeline and track progress

## Methods
Describe the detailed steps necessary to perform the protocol, including any safety precautions, data collection procedures, or data analysis methods. It should be written in a clear and concise manner to ensure reproducibility.

### Sub-points:
- Define the data collection process and data analysis methods
- Identify any safety precautions that need to be taken during the development process
- Create a detailed step-by-step plan for the development process
- Develop a testing plan to ensure the software meets all requirements

## Results
Specify what results are expected to be obtained by the protocol. Include any expected outcome, data interpretation or statistical analysis.

### Sub-points:
- Define the criteria for success
- Identify the expected outcomes of the protocol
- Develop a plan for interpreting and analyzing the data collected during the development process

## Discussion
Discuss any limitations, alternative methods or any observations made during the performance of the protocol. Also, mention any future directions for improvement or modification of the protocol.

### Sub-points:
- Discuss any limitations or challenges encountered during the development process
- Identify any alternative methods that could have been used to achieve the objective
- Discuss potential future directions for improving or modifying the software development process

## Conclusion
Summarize the overall purpose and significance of the protocol, including any implications for future research or practice.

## References
List all the references cited in the protocol.

## Appendix
Include any supplementary materials that may be relevant to the protocol, such as diagrams, flowcharts, tables, or additional data.

Das sieht super aus! Reichhaltiger, substanzieller und sauberer als das, was wir in den anderen Artikel-Teilen erzeugt haben. Nicht nur das Endergebnis ist sauberer, auch der Weg dahin. Und das hängt unmittelbar zusammen. Verstehen wir also die Notwenigkeit von sauberen Iterationen und achten auf diesen Aspekt, erreichen wir auch mit weit höherer Wahrscheinlichkeit unser Ziel. Der Preis den wir eingangs zahlen ist, dass wir weniger Relevanz zu uns, also “maexware solution GmbH” haben. Das stört aber nicht, denn das, was wir jetzt haben ist sehr umfassend in Struktur, lässt sich fast überall wiederverwenden als Markup und wenn dann das nächste Meeting ansteht muss ohnehin jemand in den sauren Apfel beissen und das konkrete Meeting, was ansteht, fein säuberlich protokollieren - zumindest muss derjenige dann nicht mehr darüber nachdenken, in welcher Struktur und in welcher Form!

Wichtiger Hinweis: In unseren letzten Versuch haben wir drei frische, voneinander unabhängige, Test-Konversationen gestartet, alle mit denselben Iterations-Requests. Die Responses waren bei allen drei Konversationen nahezu identisch. Sie unterschieden sich in Satzform, Satzbau, verwendete Wörter, aber nicht essenziell im Ergebnis. Bei einem von diesen drei Versuchen hatte ChatGPT einmal, statt Markup in einer Code-Box auszugeben, Prosa verfasst und wir sahen innerhalb der ChatGPT-GUI mit Markup formatierten Text. Bei diesem Versuch sagten wir ChatGPT:

“Provide the last proposed Template in markup in a Code-Box, so i can copy it as Markup.”

Und auch diese Konversation kam dann zum nahezu selben Ergebnis.

Unsere Learnings:

  • Das Grundkonzept einer jeden Anfrage und Antwort an und von ChatGPT ist “die Iteration”. Anders geht ChatGPT nicht vor. Es erinnert nicht, stattdessen berücksichtigt es “lediglich” vorangegangene Iterationen derselben Konversation für den Ausgang der aktuelle Iteration. Das ist ein wichtiger und qualitativer Unterschied. Wir Menschen sind nicht nur in der Lage Vorangegangenes eines Gesprächs zu erinnern und zu berücksichtigen, sondern können auch instinktiv entscheiden, welche Teil-Aspekte davon wir aus- oder einschließen wollen für unsere nächste Äußerung - wie immer wir es für richtig halten und ohne zu werten ob es dann optimal ist oder nicht. Das ist eine großartige Eigenschaft von uns Menschen, die es uns erlaubte und weiterhin erlauben wird zu überleben. ChatGPT kümmert das alles hingegen absolut nicht und konzentriert sich immer voll und ganz auf die aktuelle Iteration und die Konversation in welcher diese stattfindet, was ebenso großartig ist.
  • Wenn wir eine Konversation mit ChatGPT führen, muss uns klar sein, dass jede Referenz, Erwähnung oder “Aufgabenstellung” das Spektrum aller möglichen Antworten von ChatGPT limitiert. Das erscheint im ersten Augenblick auch logisch, da wir ja kein Liedtext haben möchten, obwohl wir im Kopf bereits wissen ein Rezept für einen Kuchen zu erfragen. Aber das Spektrum vom erlernten Wissen des Sprachmodells minimal zu limitieren, sollte immer Teil unserer Anfrage-Überlegungen sein. Setzt man ChatGPT keine Parameter, bzw. belässt den Kontext so “sauber” wie möglich, kann man großwartige Ergebnisse erwarten, bzw, zusammen mit ChatGPT erarbeiten.

Hiermit sind wir am Ende unserer “Ersten Versuche” mit ChatGPT angelangt. Ziehen wir ein kleines Resumee:

Wir haben versucht etwas strukturierter zu analysieren, was der Großteil aller menschlichen Nutzer, welche sich zum ersten Mal mit ChatGPT befassen, instinktiv machen und erfahren würde. Was vielleicht nicht jeder instinktiv macht, ist zu versuchen, die Grundprinzipien dieses neuen Anwendungs-Feldes qualitativ zu beschreiben bzw. zu schärfen, so dass man einige Grundpinzipien erkennen kann und auch Grund-Richtlinien, welche es einfacher machen dieses neue Werkzeug so zielgerichtet und effizient wie möglich einsetzen zu können. Das haben schon einige versucht, zum Zeitpunkt des Artikels und werden weiterhin noch viele versuchen. Wir geben hier unseren Anteil mit in den Pool.

In den nächsten Jahren werden diese Grundprinzipien und Richtlinien sich immer weiter schärfen und es wird eine Zeit kommen, in welcher sich Konventionen entwickeln, auf die sich große Teile der Community einigen werden. Wir werden die Zusammenarbeit mit künstlichen Intelligenzen erlernen und besser darin werden. ChatGPT, als machtvoll trainiertes Sprachmodell, ist ein erster, in der Breite präsenter Vertreter dieser künstlichen Intelligenzen.

Bis dahin, und sich solche Konventionen gebildet haben, hier unser Vorschlag und Beitrag dazu, was gleichzeitig nochmal all unsere Learnings so knapp wie möglich zusammenfasst, mit all dem verarbeitend, was wir bisher getan und erlebt haben in dieser Artikel-Reihe:

  1. Das Prinzip der Kohärenz.
    1. Maximale Kohärenz der Anfragen zur Aufgabenstellung.
    2. Anfragen kurz, knapp und so eindeutig wie möglich setzen - zur Not mehrere Iterationen einplanen.
    3. Zusatzinformationen zum Kontext nur hinzufügen, wenn ChatGPT auffordert oder ansonsten keine vernünftige Iteration erzeugen kann.
    4. Keine zu komplexe und verschachtelte Anfragen formulieren - nicht gierig sein.
  2. Das Prizip der Iteration:
    1. Eine Anfrage führt zu einer Erwiederung von ChatGPT zu ebendieser Anfrage - damit ist eine Iteration abgeschlossen.
    2. Eine Konversation mit ChatGPT ist eine Ansammlung von aufeinander folgenden Iterationen.
    3. ChatGPT wird immer die aktuelle Iteration beantworten, nicht die Konversation, in welcher diese stattfindet.
    4. Vorangegangene Iterationen werden für jede aktuelle Iteration berücksichtigt.
  3. Das Prinzip der Schärfe:
    1. Keine ungewollten Limitierungen: Vermeide in Anfragen Bezüge oder Referenzen auf einengende Sachlagen oder Subjekte, die nicht Teil der Aufgabenstellung sind, wenn das Ergebnis allgemeingültig sein soll oder der maximale Zugriff auf ChatGPTs trainiertes Wissen erhalten bleiben soll.
    2. Auflösung von Limitierungen: Konkrete Anfragen, dass bestimmte Bereiche oder Aspekte mit aufgenommen werden sollen.
    3. Strikte gewollte Limitierungen: Kontrollierter Ausschluss von bestimmten Bereichen oder Feldern, um die Erfolgsschance auf spezifische Antworten zu erhöhen.
    4. Konkrete Umwandlung einer Limitierung: Das Ergebnis, was durch Limitierung bestimmer Parameter aufkam, nutzen, um das Ergebnis für ein völlig neues Feld, was bisher ausgeschlossen wurde, zu erzeugen. Hinweis: Eine neue Konversation, welche ein Ergebnis aus einer anderen Konversation benutzt, könnte ein valider Weg sein.
  4. Das Prinzip des Kontextes:
    1. Ein Kontext ergibt sich aus allen Iterationen einer Konversation.
    2. Der sich, mit jeder Iteration weiter, entwickelnde Kontext bestimmt den Ausgang jeder einzelnen Iteration und somit der gesamten Konversation.
    3. ChatGPT hat keinen “Kontext-Clearer” oder “Kontext-Garbage-Collector”, bzw. trifft keine Entscheidungen darüber, welche Aspekte vorangegangener Iterationen verworfen werden sollen und welche nicht. Hinweis: Dies sollte auch nicht die Aufgabe einer künstlichen Intelligenz, bzw. eines trainierten Sprachmodells sein - hier betreten wir die ethische Seite des gesamten Themenfeldes.
    4. ChatGPT trifft keine Entscheidungen darüber, in welche Richtung sich eine Konversation bewegen sollte oder müsste, um eine Lösung zu einer bestimmten Aufgabenstellung zu erzeugen. Hinweis: Dies sollte auch nicht die Aufgabe einer künstlichen Intelligenz, bzw. eines trainierten Sprachmodells sein - hier betreten wir die ethische Seite des gesamten Themenfeldes.
    5. Korrektur oder Verbesserungsversuche, eine fehlentwickelte Konversaton wieder in die richtige Richtung zu bringen ändert nichts an dem Vorhandensein derjenigen Iterationen, die bereits Teils des Kontextes sind. Selbst wenn die Konversation “repariert” zu sein scheint, verbleiben auch die unerwünschten Iterationen Teil der Konversation und somit des Kontextes.

Ausblick auf zukünftige Artikel:

Wir werden uns bei der maexware solution GmbH auch mit der konkreten Einbindung von ChatGPT in echte Betriebsprozesse befassen. Wir sehen die Bereiche, die oft mit dem Wort “DevOPS” erschlagen werden, als ebendie Bereiche in denen ChatGPT als Sprachmodell in der Lage sein könnte echte Mehrwerte für Unternehmen zu schaffen. Es bleibt also spannend und wir werden darüber berichten, wie dies vorangeht.

maexware solutions