Discussion about this post

User's avatar
Jan's avatar

Toejuigen of in toom houden? Het is denk ik...én in toom houden.

The famous quote;

"A computer can never be held accountable, therefore a computer must never make a management decision".

Originated from IBM's training materials in 1979. It highlights the historical belief that, given computers cannot be held responsible, decision-making power should remain with human managers. This quote resurfaces in modern discussions about AI and machine learning, sparking debates about the ethical implications of automating decision-making processes.

Here's a breakdown of the quote's meaning and significance:

Accountability:

The core idea is that computers are not legally or morally responsible for their actions. This contrasts with human beings, who can be held accountable for their decisions and their consequences.

Decision-Making:

The quote suggests that because computers lack accountability, they should not be entrusted with tasks that require judgment or could lead to significant consequences, particularly in management.

Historical Context:

The quote was created in a time when AI was in its early stages, and there was less understanding of its potential and limitations.

Modern Relevance:

As AI and automation become more prevalent in various fields, the question of accountability for AI-driven decisions is increasingly important. This has led to discussions about how to ensure responsible AI practices and how to address potential biases or errors in AI systems.

Current Challenges:

The IBM quote raises a critical point about the potential for AI to exacerbate existing biases or make decisions that are unfair or harmful. It highlights the need for careful consideration of the ethics and legal implications of AI decision-making.

Expand full comment
Michiel's avatar

Ik heb het even aan Claude Sonnet 4 gevraagd.

"Mag een AI systeem hypotheekaanvragen beoordelen?

In Nederland mogen AI-systemen hypotheekaanvragen beoordelen, maar er gelden wel strikte regels en beperkingen.

Wat is toegestaan:

AI mag worden gebruikt als ondersteuning bij hypotheekbeoordelingen, bijvoorbeeld voor het analyseren van financiële gegevens, risico-inschatting en het maken van voorlopige beoordelingen.

Belangrijke vereisten:

Menselijke controle: Een mens moet altijd de uiteindelijke beslissing nemen of kunnen ingrijpen

Transparantie: Consumenten hebben recht op uitleg over hoe de beslissing tot stand is gekomen

Non-discriminatie: Het systeem mag niet discrimineren op basis van beschermde kenmerken zoals geslacht, etniciteit of religie

Nauwkeurigheid: De AI moet betrouwbaar en accuraat zijn

Relevante wetgeving:

De AVG (privacywetgeving) geeft consumenten rechten bij geautomatiseerde besluitvorming

De Wet gelijke behandeling verbiedt discriminatie

De aankomende EU AI Act zal extra eisen stellen aan AI-systemen in de financiële sector

In de praktijk:

Veel Nederlandse banken gebruiken al AI-ondersteuning bij hypotheekaanvragen, maar altijd met menselijke supervisie. Complete automatisering zonder menselijke betrokkenheid is problematisch vanwege de complexiteit van hypotheekbesluiten en de impact op consumenten.

Bent u geïnteresseerd in een specifiek aspect van AI bij hypotheken, zoals de technische implementatie of de juridische aspecten?"

Expand full comment
1 more comment...

No posts