MMarketing Against The Grain
← All episodes
23 February 2023

Bing A.I. Search is Way More Advanced Than We Realized…

2Frameworks
12Insights

Frameworks in this episode

Self-Mastery5 steps

AI Before-and-After Learning Loop

Use AI to apply expert teachings, compare revisions, and expose the lesson

Read →
Strategy6 steps

Milestone-Triggered Adaptation

Prepare for platform shifts, then act as soon as uncertainty becomes actionable

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster12:30

AI Recommendations Are Not Necessarily Neutral or Explainable

Personalized AI recommendations may look authoritative while hiding the signals that favored one product over another. As models grow more complex, marketers and users may be unable to determine why a particular company received visibility, potentially creating an uneven market.

  • Model builders may not understand how a final answer was produced
  • Users could take opaque recommendations at face value
  • A small set of products may receive entrenched advantages
  • Platform owners may have conflicts involving their own products

“So now you have a recommendation engine baked into AI and you don't know how it's favoring some products over others.”

Kieran Flanagan · 13:00

“Does it make the playing field pretty uneven?”

Kieran Flanagan · 13:00
#algorithmic bias#transparency#competition#recommendations

Hot Take· 3

Hot Take09:30

AI-Generated People Could Dominate Social Content

The hosts examine an entirely AI-generated model named Alice and consider how synthetic people will affect social platforms, dating apps, scams, and marketing. They predict that AI-generated material could become the majority of online content.

  • Synthetic people are already visually convincing
  • Dating and phishing scams are immediate misuse cases
  • AI content may become the majority on social platforms

“Alice is a model that is 100% AI generated.”

Kipp Bodnar · 10:00

“All content.”

Kieran Flanagan · 10:30
#synthetic media#social media#scams#generative ai
Hot Take20:00

Microsoft Is Buying Attention With Speed While Google Protects Search

Microsoft's aggressive Bing AI rollout generated extensive attention even when coverage focused on strange behavior. Google moved more cautiously, reflecting its greater need to protect an established search and advertising business.

  • Negative Bing coverage still attracted users
  • Microsoft has incentives to disrupt the existing market
  • Google has incentives to protect its advertising business
  • Speed can create adoption despite an imperfect launch

“I don't think any of that press is bad press.”

Kieran Flanagan · 20:30

“I think speed matters.”

Kieran Flanagan · 21:00
#microsoft#google#competition#product launch
Hot Take24:30

Sponsored Prompts Could Reverse the Search Advertising Model

Instead of purchasing placement after users enter a query, marketers may pay to influence which prompts users choose. Brands would promote questions that cause an AI assistant to surface their products favorably, reversing the current search-advertising relationship.

  • Prompt writing may become a core marketing skill
  • Suggested prompts could become advertising inventory
  • Brands may pay to shape the question rather than rank for the answer
  • The model creates new manipulation and transparency concerns

“we are gonna purchase the right to recommend what they should search for to find their product or services”

Kipp Bodnar · 25:30

“Right, make them ask the question that we appear for.”

Kieran Flanagan · 25:30
#prompt engineering#advertising#paid search#marketing

Explainer· 6

Explainer01:30

Why AI Search Is Becoming an Assistant, Not a Better Search Box

The hosts argue that conversational AI is a new category rather than a simple upgrade to traditional search. It can synthesize information, perform tasks, and interact in ways that feel more like an assistant than a retrieval engine.

  • AI chat performs work instead of only returning links
  • Conversational interactions can feel highly personal
  • The assistant model differs fundamentally from conventional search

“This is actually something new.”

Kieran Flanagan · 02:30

“It's a completely different thing.”

Kipp Bodnar · 02:00
#ai search#assistants#bing#search
Explainer04:00

The Logged-In Future of Personalized Search

Future search experiences may use authenticated interactions and extensive personal or business data to customize answers. This creates powerful personalization and advertising possibilities while increasing privacy and profiling concerns.

  • Search is expected to become a logged-in experience
  • AI systems may combine chat history with external user data
  • Personal data could enable highly specific advertising

“The search experience in the future is going to be a 100% logged in experience.”

Kipp Bodnar · 05:00

“So the data that AI will have about you, is actually gonna be pretty incredible for better or for worse.”

Kieran Flanagan · 04:30
#personalization#privacy#advertising#identity
Explainer07:00

AI Moves Search From Information Retrieval to Finished Work

Traditional search gives people sources and leaves them to synthesize the material. AI assistants move extraction, organization, synthesis, and initial execution into the system itself, allowing users to request outcomes such as a completed SWOT analysis.

  • Traditional search delegates synthesis to the user
  • AI can collect information and produce a first draft
  • Users can iterate on completed work instead of assembling sources manually

“Just go do it for me.”

Kieran Flanagan · 07:30

“all that thinking and synthesis and information organization that used to be done by people, it's now being done by AI.”

Kipp Bodnar · 08:00
#synthesis#automation#swot#productivity
Explainer11:00

When AI Becomes the Peer Recommending Your Next CRM

AI recommendations could evolve from broad product comparisons into advice tailored to a company's location, size, business model, and customer base. The result resembles peer advice, except the trusted peer is an AI system trained on public and private context.

  • Recommendations can incorporate company-specific details
  • AI may replace some human peer recommendations
  • Personalization makes this different from current search

“That is like a new form of peer recommendation where the peer becomes AI versus an actual human peer.”

Kipp Bodnar · 12:00

“And that is unlike search that exists today in any way shape or form, right?”

Kipp Bodnar · 12:00
#recommendations#crm#personalization#b2b
Explainer13:30

Why Google v. Gonzalez Could Reshape Recommendation Algorithms

The episode connects Google v. Gonzalez and Section 230 to the future governance of AI. The core issue is whether algorithmic recommendations are neutral distribution or editorial decisions that make platforms responsible for what they promote.

  • Section 230 historically shields platforms from liability for user content
  • The case questions whether recommendation is an editorial act
  • A broad liability shift could disrupt search and user-generated platforms
  • The precedent may affect future AI products

“a recommendation algorithm is an editorial decision, not just like passing through of some content that somebody created.”

Kipp Bodnar · 15:00

“It totally breaks the internet.”

Kipp Bodnar · 17:00
#section 230#regulation#algorithms#liability
Explainer23:00

The Cost Gap That Slows AI Search Adoption

The hosts argue that AI chat queries are dramatically more expensive than conventional search queries. Those economics limit deployment to large technology companies and help explain cautious access controls and gradual rollouts.

  • AI chat costs far more per query than traditional search
  • Large-scale deployment creates major operating expenses
  • Only the largest technology companies can initially absorb the costs
  • Economic constraints will slow the transition

“It costs Microsoft to run a new Bing chat query, 30 cents, right?”

Kipp Bodnar · 23:30

“the economics are unfeasible at scale for most companies”

Kipp Bodnar · 24:00
#inference costs#search economics#scaling#big tech

Tool· 1

Tool17:30

Could a Content Slider Put Algorithmic Choice Back on Users?

The hosts discuss proposals for public or user-selectable algorithms but question whether complex ranking systems can realistically be exposed in a meaningful way. A simpler option is a user-controlled slider that adjusts the desired level of moderate or extreme content.

  • Publishing full algorithms may not create practical transparency
  • Modern ranking systems can be too complex for their creators to explain
  • Users could choose content-intensity settings
  • User choice may shift some responsibility away from platforms

“Well I think they're oversimplifying how complicated these algorithms are.”

Kieran Flanagan · 18:00

“And you can say, "Well I want like moderate content, I want extreme content, I want whatever kind of content."”

Kieran Flanagan · 18:30
#content moderation#algorithms#user control#transparency

Takeaway· 1

Takeaway26:00

Blue Links May Fade as Personalized AI Answers Take Over

The hosts expect conversational answers to reduce the need for traditional blue-link results, although they see this as a multi-year transition. Search will become more timely, personalized, and informed by users' own data while generative AI transforms the wider content ecosystem.

  • Traditional blue links may decline over the long term
  • Personal data will shape individualized answers
  • Prompting will matter more than keyword querying
  • Generative AI may produce most future content

“I think through blue links is going to go away with that.”

Kipp Bodnar · 26:00

“AI content is going to be the majority of content and generative AI is going to transform how we all do marketing.”

Kipp Bodnar · 27:30
#future of search#blue links#personalization#generative ai