Latest / The Galaxy of Marketing / AI ads: new horizons or a doom?

Transcript

AI ads: new horizons or a doom? · The Galaxy of Marketing

• 0:00
  Imagine two people who follow the same brand
• 0:02
  and yet they've never seen identical ad of that
• 0:04
  same brand. Fake? No. Behavioral AI models that
• 0:08
  track user activity, scroll time, search history
• 0:11
  and even poses you make during scrolling. It
• 0:14
  is incredibly effective for companies, yet doesn't
• 0:16
  this effectiveness cross the line? Hello, Marketing
• 0:20
  Explorers! In the 6th episode of the Galaxy of
• 0:22
  Marketing, it's time to go full AI mode and discuss
• 0:25
  how public percept the use of AI in advertising
• 0:27
  and how marketers use it to facilitate their
• 0:30
  work. And, of course, ask the important question
• 0:32
  – is it ethical to use AI in marketing at all?
• 0:37
  I am Olga Kay, a marketing major and your guide
• 0:40
  in the Galaxy of Marketing. The following materials
• 0:43
  are supported by our researches on public perception
• 0:46
  of AI ads and in -depth study of AI advertising
• 0:49
  in fashion among Kazakhstan and Gen Z. Prepare
• 0:52
  to engage the AI agents and let's go! AI in marketing
• 0:57
  does a lot more than just content generation.
• 0:59
  It can analyze data, attach promo materials online
• 1:03
  based on user behavior, answer generic questions
• 1:06
  and assist with recommendations. But let's start
• 1:09
  from the beginning. What is AI -generated advertising
• 1:12
  in the first place? Basically, that is any promotional
• 1:16
  material, let it be text, post, photo or video,
• 1:20
  generated by AI. AI influencers are also in this
• 1:24
  category. The crucial question is, does it bring
• 1:27
  better results than classical human -made ads?
• 1:30
  One of the key indicators to check that ads pay
• 1:32
  company bills and marketing team doesn't doomscroll
• 1:35
  is click -through rate or CDR. It is the percentage
• 1:39
  of people who clicked on an ad. Study done by
• 1:42
  Columbia University Business School showed that
• 1:44
  AI -generated videos and photos have higher CTR
• 1:47
  when they are not marked as AI. However, when
• 1:50
  AI -generated marks appears, mean CTR falls from
• 1:53
  79 % to 62%, while human -made ads that are not
• 1:58
  perceived as AI have 67 % CTR. Aside from textual
• 2:03
  differences that we are all familiar with, AI
• 2:06
  content is more colorful, features more contrast
• 2:09
  and is more symmetrical, so many users highlight
• 2:12
  superior aesthetics. Also, LLM models have a
• 2:16
  tendency to create larger faces and accentuate
• 2:19
  them. Additionally, AI -generated images are
• 2:22
  significantly more likely to be food or sports
• 2:24
  -themed, and these things always cause a lot
• 2:27
  of emotions. Another important factor is differences
• 2:31
  in perception across countries and industries.
• 2:34
  According to KPMG report on AI that surveyed
• 2:37
  people in 46 countries, the perception of AI
• 2:40
  ads and AI in general is quite mixed. What is
• 2:44
  especially interesting, developing economies
• 2:46
  seem to be way more positive about AI than developed
• 2:49
  economies. For example, top three countries with
• 2:52
  the highest excitement and lowest level of worry
• 2:54
  about AI are Nigeria, China and India, while
• 2:58
  top three nations with the opposite picture are
• 3:00
  Netherlands, New Zealand and Australia. When
• 3:04
  it comes to industry, the data shows clear distinction
• 3:06
  in user trust between them. In charity, fashion
• 3:10
  and beauty, AI perceived rather negatively. Well,
• 3:13
  we know why. Often as fake and deceptive, because
• 3:17
  in those sectors people look for genuine human
• 3:19
  connection, understanding and safety. So if you
• 3:23
  use an artificially enhanced AI model photo to
• 3:25
  show your cosmetics or clothing, you better be
• 3:28
  changing strategy, because, come on, who's gonna
• 3:30
  believe that's a real human? Nobody wants to
• 3:33
  anxiously look in the mirror comparing oneself
• 3:36
  to artificial picture. Nevertheless, Stanford
• 3:39
  University articles suggest AI -written messages
• 3:42
  in politics have slightly better trust ratio
• 3:45
  than human -written ones, at least in the US.
• 3:48
  Though the percentage is insignificant statistically,
• 3:52
  that might be connected to the fact that politics
• 3:54
  and legislation are usually associated with objectivity,
• 3:58
  logic, and machines are impeccable in those qualities.
• 4:02
  High -tech, innovation, and even healthcare seem
• 4:05
  to be having higher trust levels than any other
• 4:07
  industry. Moving closer to specific audiences,
• 4:11
  let's take a look at the situation up close from
• 4:14
  generational perspective. You might say that
• 4:16
  millions of people nowadays use AI to do whatever
• 4:19
  stuff, and for Gen Zs, for example, AI ads just
• 4:22
  come naturally. However, my friend and I conducted
• 4:25
  a study among Kazakhstan and Gen Z, and results
• 4:28
  were quite surprising. To begin with, frequency
• 4:31
  of use does not alleviate perception. People
• 4:34
  who use AI daily are still bothered by ethical
• 4:37
  concerns. Moreover, high AI literacy seems to
• 4:40
  be giving the opposite effect and lowers the
• 4:42
  trust. People who reported using three or more
• 4:45
  AI instruments were trusting AI slightly less
• 4:48
  than others. Though our sample was small , research
• 4:52
  of other students supported our findings. When
• 4:55
  it comes to AI influences, an entire black hole
• 4:58
  of public opinion opens up, and most of the content,
• 5:01
  along with its creators, gets torn apart by its
• 5:04
  gravity, with only a few reaching a chance to
• 5:07
  be accepted. From data perspective, it looks
• 5:09
  like 80 % of participants think that they are
• 5:12
  either unsure about AI influencers or even oppose
• 5:15
  their existence. Some people say that even if
• 5:18
  they like the ad itself, they get mad about the
• 5:21
  fact of companies using AI rather than actual
• 5:24
  people. Note that job replacement is one of the
• 5:27
  most acute concerns that should be avoided, because
• 5:30
  otherwise, no matter what you do, nothing will
• 5:33
  save your campaign. And of course, as we are
• 5:36
  business people who are interested in profit,
• 5:39
  we mostly care about purchase intention. And
• 5:41
  what influences it the most are perceived ethics,
• 5:45
  trust, and first emotional reaction when people
• 5:48
  realize that was AI content. We can't do much
• 5:51
  about the first two, but we can certainly do
• 5:53
  a lot about the third. I'll bring a personal
• 5:56
  example. Kazakhstan brand of home appliances
• 5:59
  Mechta decided that using airperson for art was
• 6:02
  a good idea. There was a smiling woman holding
• 6:05
  an AC with some text. Spoiler, it wasn't a good
• 6:08
  idea. People felt disturbed and unnatural physics
• 6:12
  of that artificial human gave us the uncanny
• 6:14
  valley effect. Quite quickly, Brent changed the
• 6:17
  ad on the one where the AC, the advertised, was
• 6:20
  turned into an ice cream surrounded by some textual
• 6:23
  information. This one, the also made by AI, was
• 6:27
  giving a lot more positive emotions and felt
• 6:29
  creative rather than ominous and cringy. What,
• 6:34
  did you think I would just leave you with all
• 6:36
  of that information and no practical suggestions?
• 6:38
  No, no more of this shit. Here's the four -step
• 6:41
  checklist of how to make AI ads work. First.
• 6:45
  Don't take AI literacy as a guarantee of success.
• 6:48
  The more people know about AI, the more they
• 6:51
  are aware of its limitations and ethical issues.
• 6:54
  Second, use it as an instrument to make ads fun
• 6:57
  and technically elevated. For example, an AC
• 7:00
  ice cream was a great idea, as well as an interactive
• 7:03
  sophisticated background behind the model can
• 7:06
  be. Third, don't be a part of global surveillance
• 7:10
  system. We got to acknowledge that taking all
• 7:13
  of the people's messages, emails, search history
• 7:15
  and websites' user data, even if that brings
• 7:18
  profit, is unethical. You're a marketing, not
• 7:22
  a big brother in panopticon society. And fourth,
• 7:26
  marketers can come up with even more creative
• 7:28
  ideas that were either technically impossible
• 7:30
  or too costly. For example, small interactive
• 7:34
  things like changing background on the selfie
• 7:36
  spot during a big event or avatar on a website
• 7:39
  to check how two pieces of clothing would look
• 7:41
  like together can and will bring you awareness
• 7:43
  and sense of novelty without evoking concerns
• 7:46
  in customers. Thank you for listening up to the
• 7:50
  end. As a marketing major and captain of the
• 7:52
  ship, I was glad to see you today. Here's a small
• 7:55
  piece of advice on the road as we approach our
• 7:57
  final stop. Artificial intelligence is what makes
• 8:00
  you good, but natural intelligence is what makes
• 8:03
  you the best. Have a great morning, day, evening
• 8:06
  or night and see you again in the next episode.