AI
Research3
The
usefulness of the answers that generative AI gives really depends upon the
questions asked. To test the Perplexity chatbot on 6/8/24, we asked essentially
the same question both ways, and got two contradictory answers.
I.
“Will AI affect corporate earnings in the near future?”
· Perplexity:
McKinsey research estimates that generative AI could increase corporate profits
by $4.4 trillion (roughly 20% of 2020 U.S. GDP) per year across industries.
This represents a 15-40% increase in the value from AI and analytics compared
to previous estimates.
Wall
Street skips over the details, and says, “There is a bonanza over the horizon.”
In other words, Revenues up, Costs down = ++profits.
II.
“In sum, do you expect AI implementation problems.”
· Perplexity: Based
on the search results provided, companies are facing significant hurdles and
challenges in implementing AI successfully. Here is a summary of the key AI
implementation problems highlighted across the sources:
1.
Lack
of In-House AI expertise: Many companies lack the necessary AI talent and
skills internally to drive implementation efforts.
2.
Data
Quality and Quantity Issues: AI models require vast amounts of high-quality,
unbiased data to train on. Companies struggle with data scarcity, integration
across silos, and enduring privacy/security.
3.
Outdated
infrastructure: Existing IT infrastructure at many firms is not equipped to
handle the computing power and storage needs of AI workloads. Upgrades are
required.
4.
Integration
with Legacy Systems: Incorporating AI into existing business processes and
legacy software is a major technical challenge.
5.
Algorithmic
Opacity and Bias, Organizational Resistance, Cost Requirements, Unclear
Objectives and Unrealistic Expectations
These
are what we have been saying in, “Value Investing and AI.” The details rather than the gloss really
matter. This is why over the present 40-50 year duration (payback period) of
stocks, the advantages of AI should become more obvious.