Editing History of Artificial Intelligence (section)

== <span style="color: #FFFFFF;">Analyzing</span> ==
{| class="wikitable"
|+ AI Paradigm Comparison Across History
! Era !! Dominant Approach !! Key Strength !! Key Failure || Why It Ended
|-
| 1950s–1970s || Symbolic logic/search || Elegant reasoning || Doesn't scale, no robustness || Real-world complexity exceeded hand-coding capacity
|-
| 1980s || Expert systems || Domain expertise encoded || Brittle, can't learn || Maintenance cost, narrow applicability
|-
| 1990s–2000s || Statistical ML || Data-driven, generalizable || Requires feature engineering || Deep learning automated features
|-
| 2012–present || Deep learning || Learns from raw data || Needs lots of data + compute || (Ongoing — challenges remain)
|}

'''Recurring failure modes in AI history''': Overpromising on timelines (strong AI "10 years away" repeatedly). Brittle systems that work in the lab but fail in deployment. Neglecting negative results — only successes are published. Anthropomorphizing AI capabilities based on surface impressions. The "Eliza effect" — humans projecting more intelligence onto AI than is present.
</div>

<div style="background-color: #483D8B; color: #FFFFFF; padding: 20px; border-radius: 8px; margin-bottom: 15px;">